Sie sind auf Seite 1von 553

The Effects of Competition

This page intentionally left blank


The Effects of Competition

Cartel Policy and the


Evolution of Strategy and
Structure in British Industry

George Symeonidis

The MIT Press


Cambridge, Massachusetts
London, England
( 2002 Massachusetts Institute of Technology

All rights reserved. No part of this book may be reproduced in any form by any elec-
tronic or mechanical means (including photocopying, recording, or information storage
and retrieval) without permission in writing from the publisher.

This book was set in Palatino in ‘3B2’ by Asco Typesetters, Hong Kong and was
printed and bound in the United States of America.

Library of Congress Cataloging-in-Publication Data

Symeonidis, George.
The effects of competition : cartel policy and the evolution of strategy and structure in
British industry / George Symeonidis.
p. cm.
Includes bibliographical references and index.
ISBN 0-262-19468-6 (hc. : alk. paper)
1. Corporations—Great Britain—History—20th century. 2. Cartels—Great
Britain—History—20th century. 3. Strategic alliances (Business)—Great
Britain—History—20th century. 4. Competition—Great Britain—History—20th
century. 5. Big business—Great Britain—History—20th century. 6. Industrial policy
—Great Britain—History—20th century. 7. Prices—Great Britain—History—20th
century. 8. Costs, Industrial—Great Britain—History—20th century. I. Title.
HD2845 .S95 2002
338.8 0 7—dc21 2001044444
Contents

Acknowledgments ix

1 An Introductory Overview 1
1.1 A Natural Experiment 1
1.2 Key Theoretical Concepts 3
1.3 Empirical Methodology 7
1.4 Outline of the Book 12
1.5 Policy Implications 18

2 Cartel Policy and the Evolution of Competition in British


Industry 21
2.1 Introduction 21
2.2 The Origins of British Cartel Policy 22
2.3 The 1956 Act 23
2.4 Restrictive Practices in the 1950s 33
2.5 The Impact of the 1956 Legislation on Firm Conduct 40
2.6 Concluding Remarks 43

3 The Competition Data 45


3.1 Introduction 45
3.2 Data Sources on Competition in British Industry 46
3.3 The Construction of the Data Set 56
3.4 The Determinants of Collusion 71
3.5 Concluding Remarks 83

4 Price Competition and the Evolution of Concentration in


Exogenous Sunk Cost Industries 87
4.1 Introduction 87
4.2 Theoretical Framework 91
vi

4.3 A Specific Example: The Linear Demand Model 100


4.4 The Data 104
4.5 Empirical Model and Results 124
4.6 Concluding Remarks 139

5 Price Competition, Advertising, and Market Structure in


Advertising-Intensive Industries 149
5.1 Introduction 149
5.2 Theoretical Framework 151
5.3 A Specific Example: The Linear Demand Model with
Quality 164
5.4 The Data 172
5.5 Econometric Models and Results 178
5.6 Two Case Studies 211
5.7 Concluding Remarks 221

6 Price Competition, Innovation, and Market Structure in


R&D-Intensive Industries 223
6.1 Introduction 223
6.2 Theoretical Framework 229
6.3 The Data 238
6.4 Empirical Models and Results 250
6.5 Concluding Remarks 274

7 Price Competition and Profitability: Are Cartel Laws Bad for


Business? 277
7.1 Introduction 277
7.2 Theoretical Issues 279
7.3 Case-Study Evidence 285
7.4 The Data 294
7.5 Empirical Models and Results 302
7.6 Concluding Remarks 323

Epilogue 327

Appendix A: A Survey of Collusive Agreements in British


Manufacturing Industries 333

Appendix B: Data Sets 415


Table B1: Data set for chapter 4 416
Contents vii

Table B2: Data set for the concentration regressions of


chapter 5 432
Table B3: Data set for the advertising regressions of
chapter 5 441
Table B4: Data set for the innovation regressions of
chapter 6 455
Table B5: Data set for the concentration regressions of
chapter 6 465
Table B6: Data set for chapter 7 474

Glossary of British Terms 509


References 511
Index 523
This page intentionally left blank
Acknowledgments

The research for this book began several years ago, when I was a
graduate student at the London School of Economics, and continued
after I took my current post at the University of Essex. My intellec-
tual debt to John Sutton is almost too obvious to be acknowledged.
His invaluable advice in the course of many and long discussions
has profoundly influenced my work, and I am also grateful to him for
numerous insightful comments on various drafts, including those on
the penultimate draft of this book.
I have also benefited greatly from discussions with several other
people on various aspects of the research presented here. I am espe-
cially obliged to V. Bhaskar, Paul Geroski, Tim Hatton, Sajal Lahiri,
Stephen Machin, Pedro Marin, Mark Schankerman, Philip Vergau-
wen, Michael Waterson, and Hugh Wills for their very helpful advice
and suggestions at various stages of the project. I also wish to
thank seminar audiences at the London Business School; the London
School of Economics; Nuffield College (Oxford); the Universities of
Athens, Carlos III (Madrid), East Anglia, Essex, Leuven, Maastricht,
and Warwick; and at numerous conferences, where parts of this re-
search were presented over the past few years.
I owe a very particular debt to David Genesove and Stephen
Martin, both of whom read the entire manuscript and provided very
detailed and thoughtful comments. Their suggestions led to signifi-
cant improvements in the final version.
The financial assistance of the Greek State Scholarships Founda-
tion helped me begin the research reported in this book. Some of this
work has appeared in different form in a number of journal articles,
and I wish to thank the publishers of Economica (Blackwell Publishers
and London School of Economics), the Journal of Industrial Economics
(Blackwell Publishers), and the Journal of Economics and Manage-
x

ment Strategy (MIT Press) for permission to use material from these
sources. I am indebted to Terry Vaughn, former economics editor at
the MIT Press, for his help and encouragement, and I am also thankful
to John Covell for dealing very efficiently with the manuscript.
Finally, my warmest thanks go to my parents, Antonis and Niki,
and to my wife, Angelina, for their continuous support and encour-
agement over many years.
1 An Introductory Overview

1.1 A Natural Experiment

This book is a study of the effects of competition on firm strategy,


market structure, and industry performance. It combines game-
theoretic models of competition with econometric and case-study
evidence from a unique natural experiment that occurred in the UK
between the late 1950s and the early 1970s. The introduction of the
1956 Restrictive Trade Practices Act led to the registration, and sub-
sequent abolition, of explicit restrictive agreements between firms
and the intensification of price competition across a range of manu-
facturing industries. An equally large number of industries had not
been subject to agreements significantly restricting competition, and
were therefore not affected by the legislation. A comparison of the
two groups of industries over a twenty-year period, using data from
both before and after the introduction of the 1956 Act, can provide
important insights on the effect of price competition on concentra-
tion, firm and plant numbers, profitability, advertising intensity, and
innovation.
The introduction of cartel policy in the UK is a very rare oppor-
tunity for a systematic and comprehensive empirical analysis of the
effects of competition. The most important feature of this natural ex-
periment is that it provides us with a way to bypass two very dif-
ficult problems that have been endemic in empirical studies of the
effects of competition. The first problem is how to measure the in-
tensity of competition. The second problem is how to unravel the
complex links between competition and other variables, such as
market structure, innovation, and profitability, given that all these
variables may simultaneously affect one another, thus making the
identification of one-way causal effects very difficult.
2 Chapter 1

The present setup allows us to bypass these difficulties because a


change in the intensity of competition across a wide range of indus-
tries was in this case induced by an exogenous and measurable
institutional change. Thus there is no need to measure the intensity
of competition directly. All that is required is a clear distinction be-
tween industries affected by the shift in cartel policy and industries
not affected. Moreover, the exogeneity of the institutional change
allows us to largely overcome any concerns about potential biases in
the estimated impact of competition caused by the existence of com-
plex links between competition and other variables. In other words,
the identification of one-way causal effects is a feasible task in the
present context.
Several of the issues addressed in this book have been the subject
of long-standing debates among economists as well as among policy
makers. The present research aims to contribute to a better under-
standing of these issues and to raise some additional ones. In which
industries is collusion more likely? Do cartels raise firms’ profits in
the long run and do they restrict entry into industries? What is the
effect of cartel laws on market structure and profitability across dif-
ferent classes of industries? Is there a link between recent policy
changes that have introduced more competition across many national
and international markets, and the observed restructuring of a large
number of industries? Does competition promote innovation? Do
firms advertise less, or do they advertise more, when price compe-
tition is tougher? Is higher concentration necessarily associated with
higher prices and profits? Finally, what are the constraints on the
exercise of merger and antitrust policies? This book examines these
issues, and others, using game-theoretic models of oligopoly and a
rich data set covering the whole of the UK manufacturing industry.
At a more general level, the analysis of the effects of competition in
this book aims to contribute to the recent renaissance of a research
program in industrial economics that involves formulating general
theoretical predictions and testing them to reveal empirical regu-
larities across industries. This renaissance has come as a response to
a rather pessimistic view of the robustness of theoretical and empir-
ical results in industrial economics. According to that view, since the
results of game-theoretic models of oligopoly often depend on spe-
cific assumptions about a range of factors (some of which are impos-
sible to identify empirically) and the observed empirical outcomes
are partly driven by particular and nonsystematic characteristics
An Introductory Overview 3

of industries, ‘‘anything can happen in oligopoly.’’ In this book I


argue, in line with recent theoretical and empirical work on the de-
terminants of market structure (see Sutton 1991, 1998), that this view
is unnecessarily pessimistic. In particular, I analyze a small number
of mechanisms, involving observable variables, which operate in a
systematic way across industries or for a broad class of industries.
Accordingly, the theoretical predictions developed in this book are
tested using cross-industry data.

1.2 Key Theoretical Concepts

The empirical analysis of the effects of the 1956 Restrictive Trade


Practices Act in this book is firmly based on game-theoretic models
in which the intensity of price competition is exogenously deter-
mined, while market structure is endogenously determined by
means of a free-entry condition. A central feature of these models is
the assumption that in long-run equilibrium, net profit is restricted
by free entry, irrespective of the type of competition regime facing
firms. A change in firm conduct induced by an institutional change,
such as the introduction of legislation prohibiting cartels or economic
integration, will have little effect on net profit, since it will cause
industries to move from one long-run free-entry equilibrium to an-
other. It will, however, have a significant effect on market structure,
and it could also affect innovation, nonprice competition, and the
labor market. It is precisely through adjustments in these variables,
especially market structure, that a new long-run equilibrium will be
reached.
This theoretical framework encompasses key insights of the tradi-
tional structure-conduct-performance approach, according to which
profitability is higher in more concentrated industries because con-
centration facilitates the exercise of market power by firms. In par-
ticular, the present theory allows for a negative effect of entry on
profit margins. However, it embeds this in a richer framework that
allows for a two-way link between strategy and structure. Unlike the
traditional approach, which sees market structure as being largely
determined by exogenous ‘‘barriers to entry,’’ the present theory em-
phasizes how changes in firm conduct affect the conditions of entry,
and hence market structure.
The unraveling of the two-way link between strategy and structure
is carried out by applying a simple distinction between firms’ long-
4 Chapter 1

run and short-run decisions. Long-run strategic variables include


the decision to enter an industry by setting up a plant, and also the ex-
penditures involved in inventing or developing a product, improv-
ing a production process, or creating an advertising-based brand
image; they may also include the outcome of bargaining with a
union over the division of rents. These are considered to be long-
run choices because they are relatively difficult to reverse in the
short run. On the other hand, short-term decisions are easily rever-
sible and generally involve setting the price of the product or/and
the level of output produced in any period. The precise link between
long-run and short-run decisions can be summarized as follows.
Long-run decisions determine a set of conditions that firms must
take as given when making short-run choices. At the same time, firms
anticipate, when making their long-run decisions, the way these
decisions will affect their own and their rivals’ short-run choices, and
they act accordingly.
It is firm conduct with respect to the short-run choice variables
that is referred to as ‘‘intensity of price competition’’ throughout this
book. This should not be confused with the price-cost margin, which
is an index of performance rather than of conduct. The difference
between the two is important and requires some further clarification.

The Intensity of Price Competition

The concept of the intensity of price competition is intended to cap-


ture the idea that, for any given market structure and set of long-
run variables, profit margins will depend on firms’ pricing strategies,
which in turn will partly depend on exogenous institutional factors
such as the climate of competition policy or the degree of eco-
nomic integration. Thus, for instance, a switch from collusive to non-
collusive pricing behavior represents an increase in the intensity
of price competition. On the other hand, a change in the price-cost
margin may occur even in the absence of any change in pricing con-
duct. The key point is that the intensity of price competition refers to
the type of pricing behavior, not to the level of price-cost margins. It
may, in fact, be useful to think of the intensity of price competition
as a functional relationship between a set of variables that are taken as
fixed in the short run, including market structure, and the level of
margins. A change in the intensity of price competition will then
An Introductory Overview 5

imply a shift in this functional relationship. For instance, an increase


in the intensity of price competition will imply that price-cost mar-
gins are lower at any given level of concentration.
The attractiveness of this concept for the study of competition in
oligopolistic markets stems from the fact that it bypasses the need to
specify or observe the determinants of firms’ pricing behavior, many
of which are unobservable or nonsystematic variables. It may, how-
ever, be asked whether we are justified in treating the intensity of
price competition (but not the price-cost margin) as an exogenous
variable in this book. This is not unreasonable on theoretical grounds,
given the well-known multiplicity of equilibria in dynamic game-
theoretic models of oligopolistic interaction. In these models, there is
a wide range of possible outcomes regarding pricing behavior for any
given set of parameters, ranging from perfect collusion (i.e., joint profit
maximization) to pricing at marginal cost. Moreover, and more im-
portant, I would argue that the simplifying assumption of exogeneity
of the intensity of price competition is probably justified in the
present context for two reasons. First, because cartelization in British
manufacturing in the 1950s seems to have been largely a function of
exogenous industry-specific factors rather than of endogenous vari-
ables such as concentration. Second, because changes in firms’ pricing
behavior during the 1960s were largely determined by an exogenous
institutional change, namely, the shift in cartel policy. Both points
will be discussed extensively in later chapters.

Exogenous and Endogenous Sunk Costs

Following Sutton (1991, 1998), an important distinction within the


present theoretical framework is between exogenous sunk cost in-
dustries, advertising-intensive industries, and R&D-intensive indus-
tries. The first of these categories includes industries where the only
important sunk cost is the exogenously determined cost of setting up
a plant of minimum efficient scale, net of resale value. In such in-
dustries there is limited scope, at any given time, for cost-decreasing
or quality-increasing investment. The second and third categories
include industries in which firms incur, in addition to the exogenous
setup cost, significant endogenous sunk costs, such as advertising or
R&D, in order to reduce unit costs or increase the consumers’ will-
ingness to pay by enhancing product quality or brand image. These
6 Chapter 1

costs are sunk before firms set their price or output, and they are
endogenous because they can vary depending on each firm’s own
decisions.
It is, of course, true that firms may make decisions on nonprice
variables in all industries. The concept of an exogenous sunk cost
industry is simply a useful polar case that captures the fact that
in many industries, advertising and R&D are not important stra-
tegic considerations. In other words, the general theoretical model
throughout this book is one that allows for endogenous sunk costs.
But within that model there is a special limiting case which arises
when advertising effectiveness and technological opportunity are
both very low. In this limiting case, firms spend nothing on adver-
tising or R&D. Thus the equilibria of this limiting-case model coin-
cide with the equilibria of an elementary model in which there is no
advertising or R&D, and hence the only sunk cost is the exogenously
determined cost of entry. The elementary model can be used to ana-
lyze industries with zero or very low advertising and R&D expendi-
ture. On the other hand, the general model is the appropriate model
for advertising-intensive and R&D-intensive industries.
The notion that certain costs are sunk is closely related to the dis-
tinction between firms’ long-run and short-run decisions. At any
point in time, some variables are more difficult to change than
others. For instance, the decision to enter an industry by setting up a
plant is more difficult to reverse, once implemented, than the choice
of a price for the product. The reason is that capital stock is difficult
to resell and depreciates only gradually, while price setting does not
involve any kind of irreversible investment. It follows that the entry
decision must be taken as given, and the cost of setting up a plant is
largely a sunk cost, when a firm sets a price for its product. A game
structure that captures this distinction is that of the multistage game,
with the long-run decisions being taken at the earlier stages and
regarded as given when the short-run decisions are taken in the late
stages of the game.
The simplest structure that can be used to model competition in
an exogenous sunk cost industry is a two-stage game where firms
decide whether to enter or not to enter at stage 1, then choose a price
or a level of output at stage 2. To find the equilibrium of such a
game, we have to ask how the entry decision is influenced by the
anticipated outcome of short-run price or output competition. Market
structure, that is, the number of firms that enter the market and their
An Introductory Overview 7

respective market shares, is endogenously determined as part of that


equilibrium.
On the other hand, a three-stage game—with an entry decision at
stage 1, choice of advertising and/or R&D expenditure at stage 2,
and choice of price or output at stage 3—is the simplest game struc-
ture appropriate for endogenous sunk cost industries. The procedure
for solving this three-stage game involves asking (1) how the choice
of advertising/R&D expenditure is affected by the anticipated out-
come of short-run price or output competition, and (2) how the entry
decision is influenced by the anticipated outcome of the entire sub-
game that begins at stage 2. The equilibrium now involves the de-
termination of both market structure and the level of advertising/
R&D expenditure.

1.3 Empirical Methodology

The construction of the data set for this book has involved examining
a very large number of restrictive agreements registered under the
1956 Restrictive Trade Practices Act. Other sources of information
on competition in British manufacturing industry during the 1950s,
1960s, and 1970s were also used. It has therefore been possible to
assign all manufacturing industries to one of three groups: those
affected by the legislation, those not affected, and those which, for
one or the other reason, were difficult to classify. The competition
data were matched against data on several other variables, including
concentration, firm and plant numbers, profits and profit margins,
advertising intensity, and innovation counts, obtained from a variety
of official and unofficial sources. Several different samples of indus-
tries were constructed and used to analyze a variety of empirical
issues. The ambiguous group of industries was excluded from these
samples, and the results were based on comparisons between the
group of industries with a change in competition regime and the
group without such a change.
The time period considered spans a total of more than twenty
years, from the mid-1950s to the mid-1970s. For many of the vari-
ables, data are available only at roughly five-year intervals during
most of this period. In particular, Census of Production data are avail-
able for 1954, 1958, 1963, 1968, and then annually from 1970 onward.
The years 1954 and 1958 are ‘‘before’’ dates in the natural experiment,
the former because the legislation had not yet been introduced, the
8 Chapter 1

latter because it had not yet been effectively implemented. Moreover,


since the effects of the introduction of cartel policy were gradual and
took several years to be fully realized, one should expect to see only
the short-run effects of competition by 1963. By 1968 much of the
long-run effect should have appeared, and by the early 1970s the full
long-run effects of competition had certainly been realized.
The comparisons between industries with a change in competition
regime and industries without such a change will rely, for the most
part, on the use of panel data econometrics. This has been an obvious
choice. By revealing average tendencies in the data, the econometric
analysis of the effects of the 1956 Restrictive Trade Practices Act pro-
vides results that are fairly general. Moreover, it allows us to control
for any measurable factors other than the introduction of cartel pol-
icy that had an effect on the endogenous variables of interest during
the period examined. It also allows us to control for unobserved time-
invariant, industry-specific characteristics, that is, unobservable fac-
tors that may be correlated with some of the variables of interest but
are more or less constant over time for any given industry. Control-
ling for these factors implies that they will not blur the key relation-
ships of interest. For all these reasons, most of the empirical analysis
in this book is firmly based on the estimation of econometric models.
On the other hand, the econometric analysis, by focusing on the
overall effects of competition across industries, may not always be
appropriate for examining some finer aspects of the theoretical pre-
dictions, or for documenting the operation of certain mechanisms,
or indeed for illustrating some of the limitations of the theory. Case
studies can provide this detail, and they are often a useful comple-
ment to econometric work. In this book, therefore, the econometric
analysis of the effects of competition is sometimes supplemented by
detailed case studies of particular industries.

The Key Identifying Assumption

In recent years, there has been an outburst of work in economics


using empirical evidence from natural experiments. Unlike random-
ized experiments in psychology or other fields, these studies are
based on comparisons between experimental groups and control
groups that are not randomly constructed. The most frequently used
approach is the so-called difference-in-differences method, which
consists in comparing the difference between the average change in
An Introductory Overview 9

the variable of interest in the experimental group and the average


change in the same variable in the control group. Meyer (1995) pro-
vides a critical evaluation of this approach in economics, including a
discussion of possible extensions as well as potential limitations.
Since the approach used in the present book is essentially based on
the difference-in-differences methodology—although this is applied
in the context of regression analysis in order to control more effec-
tively for industry-specific effects, time effects, and changes in other
variables—it will be useful to clarify from the beginning the limi-
tations of this methodology. Several potential ‘‘threats to validity’’
are listed in Meyer (1995). These include omitted variables, trends
in outcomes, measurement error, simultaneity or selection biases,
omitted interactions, and so on. Some of these potential problems
apply to a certain extent to all empirical studies; others are more
specific to natural experiments, but they are not relevant for the
present study or are relatively straightforward to deal with in econo-
metric work; and still others are difficult to deal with in a totally
satisfactory manner. As a result, virtually all studies of natural ex-
periments must rely on some identifying assumptions to validate the
research design being used. The present book is no exception to this.
The key identifying assumption that I make in this book is that
there are no underlying trends in any of the endogenous variables of
interest which differ between the experimental and the control group
but are unrelated to the 1956 Restrictive Trade Practices Act or any
of the other control variables used in the regressions. In other words,
I must assume that any difference in the evolution of concentration,
profitability, innovations, and so on, following the introduction of
the 1956 legislation, between industries with a change in competition
regime and industries without such a change—after controlling for
changes in a set of relevant and measurable explanatory variables
—can be attributed to the effect of the 1956 legislation and is not
the result of unobserved characteristics that differ between the two
groups and cause them to evolve in different ways during the period
under study.
Clearly, the validity of this assumption is not immediately obvious
in the present context. After all, there must have been some differ-
ence in underlying characteristics between those industries that were
cartelized in the 1950s and those that were not. The success of the
present study therefore depends crucially on being able to argue that
this difference does not invalidate the central identifying assumption
10 Chapter 1

stated above. Admittedly, as in all studies of natural experiments, it


is impossible to test the identifying assumption directly. I believe,
however, that a strong case for accepting this assumption can be
made on the basis of two sets of indirect tests.
First, as already pointed out above and discussed in considerable
detail in later chapters, cartelization in British manufacturing in the
1950s seems to have been largely a function of exogenous factors,
including time-invariant, industry-specific characteristics, rather than
endogenous variables such as concentration. Second, and more im-
portant, later chapters examine the key issue of whether trends in the
endogenous variables of interest were different between the experi-
mental group and the control group before the effective implementa-
tion of the cartel legislation. As it turns out, this evidence suggests
that the estimated effects of the intensification of price competition
following the introduction of the 1956 Restrictive Trade Practices Act
are not biased by any preexisting trends that differ between the two
groups.

Further Remarks and Clarifications

Two further remarks on the present research design and the links
between theory and empirical testing are in order. Much of the
theoretical and empirical analysis in this book abstracts from one
potentially important factor, namely, the fact that more intense price
competition may affect firms’ incentives to reduce their variable costs.
This can take the form of increased effort by workers and managers,
a relative reduction in wages and salaries, or the adoption of a better
technology.
While these are important issues, they are outside the scope of
this book. The assumption made here is that any such reduction in
variable costs will be largely passed on to prices and will have little
effect on price-cost margins. On the other hand, changes in price-
cost margins will be largely driven by changes in competition, advertis-
ing, and R&D intensity, and the number of firms. Consequently, any
potential effects of competition on the labor market and on produc-
tivity will not substantially alter the basic links between competition,
on the one hand, and market structure, advertising intensity, innova-
tion, and profitability, on the other, which are the subject of this book.
It may also be asked whether the assumption that firms are sym-
metric and produce one product each, which is made in the theoret-
An Introductory Overview 11

ical sections of the book, is in fact too restrictive or misleading as to


the key mechanisms at work at the empirical level. There are two
responses to this. The first is to modify the predictions regarding the
equilibrium levels of the endogenous variables in order to account
for the presence of asymmetric multiproduct firms. This would be
in line with Sutton’s (1991, 1998) ‘‘bounds’’ approach to the study of
market structure.
According to this approach, there is, for any given set of measur-
able and systematic exogenous factors, a multiplicity of possible
outcomes, which is largely due to the presence of asymmetric multi-
product firms. It is then of no use to try to predict a unique outcome
on the basis of the observable and systematic characteristics of an
industry; the realized equilibrium will also depend on unobservable
features of the industry as well as on its history (which is to some
extent the product of chance). The bounds approach involves placing
bounds on the space of outcomes and thus excluding certain out-
comes as impossible or unlikely, while admitting others as possible
or likely equilibria. What is the implication of this in the present con-
text? The implication is that the bounds approach would open up
the possibility that a change in the competition regime has no effect
on certain industries in the long run. This is because such a change
would be seen as shifting the bounds that the theory places on the
space of outcomes rather than the realized equilibria. Hence, while
it would certainly affect industries initially close to the bounds (be-
cause the original equilibria would be, after the change in the com-
petition regime, outside the space of possible outcomes), it might or
might not affect industries initially well inside the bounds (because
the original equilibria would then remain inside the space of possible
outcomes).
However, the bounds approach is difficult to apply in the
present context, chiefly because of the panel structure of the data.
As explained in detail in later chapters, this results in methodological
problems with the econometric estimation of bounds. Standard
panel regression techniques therefore seem to be the most appro-
priate approach in the present case.
The second response to the issue of asymmetric multiproduct firms
is to argue that the predictions regarding the equilibrium changes in
the endogenous variables are valid, at least on average, since it is
very improbable that the mechanisms identified for the case of
symmetric single-product firms would be offset across industries by
12 Chapter 1

other mechanisms working in different directions in a more complex


setting. In other words, the change in the competition regime will
trigger mechanisms that will shift the realized equilibria and will be
relevant for all industries. While it is conceivable that other influ-
ences might operate to mitigate or offset these mechanisms in some
cases, any such influences are unlikely to be systematic. In the
context of the bounds approach, this is another way of saying that
whether an industry is close to the predicted bounds or well inside
the bounds is determined by nonsystematic factors; hence, any cases
where the change in the competition regime has little or no effect
because the industry is well inside the initial bound will be balanced
by cases where the effect is disproportionately large because an in-
dustry moves from a position close to the initial bound to one well
inside the new bound. As a result, we can expect a change in the
competition regime to be of relevance for all industries, and any devi-
ations from the general pattern to be due primarily to random factors.
This second response motivates the approach adopted in this book.
Its validity must, of course, be tested against the empirical evidence.
And it is also important to highlight instances where the presence of
asymmetries between firms calls for small modifications to the theo-
retical predictions obtained from the benchmark case of symmetric
firms, which should also be reflected in the evidence.

1.4 Outline of the Book

The core of this book (chapters 4 to 7) consists of a detailed analysis


of the effects of competition on market structure, firm strategy, and
profitability, combining theory and evidence from British manu-
facturing industry. Before embarking upon this analysis, however, it
is necessary to provide some background information on the evolu-
tion of competition in British industry and to discuss the method-
ological issues involved in defining the two groups of industries:
those with a change of competition regime following the introduc-
tion of the 1956 Restrictive Trade Practices Act and those without a
change in regime.

Cartel Policy and the Evolution of Competition in British Industry

Chapter 2 describes the institutional context of the present study and


provides details on the evolution of competition in British manufac-
An Introductory Overview 13

turing industry from the early 1950s to the mid-1970s. The purpose
of this chapter is to provide the essential background information on
the natural experiment analyzed in the book. I therefore begin by
describing the structure of the 1956 legislation and its implementa-
tion. I explain, among other things, why the large majority of indus-
tries registered their restrictive agreements rather than immediately
dropping them or secretly continuing them, and why it took more
than a decade for the effects of the 1956 Act on competition in British
industry to be fully realized. Moreover, rather than assume that the
Act had a significant effect on firm conduct, I set out to justify this
claim on the basis of the available case-study evidence. The conclu-
sion is that price competition did indeed intensify, following the
introduction of restrictive practices legislation, in the large majority
of the previously cartelized industries.
A more in-depth analysis of competition and collusion in the 1950s
is carried out in chapter 3. In the first part of this chapter I look at
the state of competition across British manufacturing industries in the
1950s, and I argue that the existing evidence allows us to distinguish
three groups: (1) a group of collusive industries (defined as indus-
tries with explicit agreements significantly restricting competition),
the large majority of which subsequently experienced a change of
competition regime; (2) a group of industries without explicit collu-
sive agreements, which therefore experienced no change in regime;
and (3) a residual group of industries for which we are uncertain as
to their state of competition. I also describe in some detail the con-
struction of the competition data used throughout the book.
In the second part of the chapter I examine which structural in-
dustry characteristics were associated with the occurrence of collu-
sion in the UK in the 1950s. The purpose of this exercise is twofold.
On the one hand, it is intended as a contribution to the literature on
the factors facilitating or hindering collusion. The significance of this
question is rather obvious: attempts by firms to establish price-fixing
arrangements are a matter of great concern for public policy, since
such behavior is regarded as being against the public interest, and is
therefore illegal under most competition laws. On the other hand, the
final part of chapter 3 prepares the way for the analysis of the natural
experiment by examining the similarities as well as the differences in
initial conditions between the two groups of industries. A key result
is that the two groups were broadly similar with respect to initial
market structure, but rather different with respect to capital intensity
14 Chapter 1

or the level of entry costs, advertising effectiveness, and technological


opportunity. Thus the occurrence of collusion in British manufactur-
ing industries in the 1950s can be seen as being largely determined
by exogenous variables rather than by endogenous variables such as
concentration.

The Effect of Price Competition on Concentration, Advertising, and


Innovation

Chapter 4 begins by introducing a general theoretical framework for


the analysis of the competition-market structure relationship in ex-
ogenous sunk cost industries. One of the key results to emerge from
the recent literature on the determinants of market structure is that,
in exogenous sunk cost industries, an intensification of price compe-
tition is expected to cause a rise in concentration. This is the result of
a structural mechanism that involves a movement between two free-
entry equilibria. The intensification of price competition reduces
profit margins, given the initial level of concentration. As a result,
firms (or some firms) can no longer cover their sunk costs at the
initial free-entry equilibrium: profit margins are too low to generate
a normal rate of return on the setup cost incurred in establishing
plants. This will inevitably lead to mergers and exit until the gross
profit of each remaining firm rises sufficiently to cover sunk costs.
Chapter 4 examines this mechanism in the context of the UK man-
ufacturing industry, focusing on the evolution of concentration over
the period 1958–1975. The econometric results, based on a compari-
son of industries affected by the 1956 Restrictive Trade Practices Act
and those not affected, suggest that the intensification of price com-
petition following the termination of price-fixing agreements led to a
rise in concentration in exogenous sunk cost industries affected by
the legislation. In particular, the results indicate that cartel policy
raised, on average, the five-firm concentration ratio by six to seven
percentage points in exogenous sunk cost industries. To put this fig-
ure in some perspective, note that nearly half of the entire British
manufacturing sector was cartelized in the 1950s and the aver-
age five-firm concentration ratio across all manufacturing industries
increased by about eight percentage points between 1958 and 1975,
in what was the most significant restructuring of British industry for
several decades. The results of chapter 4 suggest that cartel policy
was one of the key factors behind this restructuring.
An Introductory Overview 15

Chapter 5 focuses on advertising-intensive industries. Things are


more complicated in this case, because the intensification of price
competition is likely to affect firms’ incentives to spend on advertis-
ing. A first question in this context relates to the link between price
and nonprice competition, that is, whether tougher price competi-
tion can have a negative or a positive effect on advertising intensity.
Moreover, if advertising falls when price competition intensifies, the
decrease in the sunk cost that each firm must cover at equilibrium
may or may not offset the fall in gross profit due to increased price
competition, and hence concentration may decrease or increase.
The first part of chapter 5 introduces a theoretical framework for
analyzing these issues, derives general theoretical results, and illus-
trates these results using a specific model. Given that a decrease in
the degree of collusion in the short-run variable can have ambigu-
ous effects on advertising and concentration in industries with high
advertising intensity, can we still derive any theoretical predictions
about the effects of competition in this class of industries? It turns
out that some restrictions can be imposed by theory on the joint
behavior of market structure and advertising expenditure following
an intensification of price competition. Thus, certain outcomes, such
as a fall in concentration together with an increase in advertising
expenditure, cannot occur, since they are inconsistent with the
requirement that net profits be zero in free-entry equilibrium. More-
over, stronger predictions on the competition-market structure rela-
tionship, which are conditional on the behavior of other observable
variables, are also derived. Additional insight on these issues is ob-
tained from the analysis of a specific theoretical model.
In the second part of chapter 5, the theoretical predictions are
tested using econometric and case-study evidence from British
advertising-intensive manufacturing industries from 1954 to 1975.
The econometric results indicate that the intensification of price com-
petition following the 1956 legislation caused, on the whole, a rise
in concentration in advertising-intensive industries and probably also
a fall in advertising intensity. The magnitude of the overall effect
on concentration was not any lower in advertising-intensive indus-
tries than in exogenous sunk cost industries. On the other hand, the
case-study evidence confirms that a fall in concentration, following
an intensification of price competition, cannot be ruled out in high-
advertising industries. However, more intense price competition
must lead to a significant fall in advertising intensity if concentration
16 Chapter 1

does not rise—or, conversely, it must cause concentration to rise if


advertising falls little or not at all. This is consistent with the theo-
retical predictions as well as with the overall picture given by the
econometric results.
The analysis of the effect of price competition on market struc-
ture and technological innovation in R&D-intensive industries is the
subject of chapter 6. The chapter begins by adapting the theoretical
framework developed in chapter 5 to the analysis of R&D-intensive
industries. In contrast with most of the previous literature on the
links between market power and innovation—but in line with the
empirical evidence on the effect of the 1956 Restrictive Trade Prac-
tices Act on market structure—I do not regard concentration as a
proxy for market power. In fact, I explicitly treat both market struc-
ture and innovation as endogenous variables, while the intensity of
price competition is taken as exogenous. The model predicts, first,
that the effect of price competition on innovation is ambiguous and,
second, that if the effect of price competition on innovation is not
negative, then its effect on concentration must be positive.
The empirical results of chapter 6 are consistent with the theory.
Using UK data on competition, concentration, and the number of
innovations between the early 1950s and the mid-1970s, I find that
the intensification of price competition following the 1956 legisla-
tion had no significant effect on firms’ innovative output and a strong
positive effect on concentration in R&D-intensive industries. The mag-
nitude of the overall effect on concentration was at least as high
as that estimated for exogenous sunk cost industries. On the other
hand, the lack of any overall effect of competition on innovation
across industries is not very surprising in light of the mixed theoret-
ical results on this much-debated issue. Perhaps a more promising
line of research in this area would be to identify specific conditions
favoring a positive or a negative effect of competition on innovation;
such an analysis could not be carried out in the present context be-
cause of data limitations. It is also interesting to contrast the absence
of any overall effect of price competition on innovations with the
likely presence of a negative effect on advertising intensity. This
raises the question of whether there is a fundamental difference be-
tween advertising and R&D in their role as nonprice choice variables
in the competitive process that has not been adequately analyzed in
existing theories of competition.
An Introductory Overview 17

Price Competition and Profitability: Are Cartel Laws Bad for Business?

Chapter 7 extends the analysis of the previous chapters by examining


the joint effect of price competition on market structure and profit-
ability. If the economic mechanism underlying the observed effect
of the 1956 legislation on British industrial structure is the one de-
scribed above, then one should also expect that there will be no
significant change in firms’ profits in the long run in exogenous sunk
cost industries—and probably also in all classes of industries, given
that price competition did not have a very strong overall effect on
nonprice variables in the present case. Moreover, one would expect
profitability to decline in the short run, that is, before any significant
change in market structure occurs, and then be restored or partially
restored in the long run through the fall in the number of firms.
Chapter 7 confirms these predictions by examining the joint evo-
lution of market structure variables, such as the number of firms and
the number of plants, and profit measures, such as the average firm
profit, the average plant profit, and the profit margin, from 1954 to
1973. Two different samples are used in the econometric analysis:
a sample pooling all classes of industries and a subsample of ex-
ogenous sunk cost industries. In both cases, the results indicate
that the 1956 legislation had a negative effect on profitability during
the first few years of its implementation, when market structure
had not yet fully adjusted. However, the subsequent restructur-
ing of industries affected by the legislation was associated with a
recovery of profit margins, so that profitability did not change much
in the long run. An interesting implication of these results is that free
entry is not incompatible with collusion. In fact, the results suggest
that in long-run equilibrium most cartels will result in excess entry
rather than excess profits (relative to the absence of collusion), and
hence cartel laws will reduce the number of firms rather than their
profits. Of course, this crucially depends on entry being unrestricted
irrespective of the intensity of price competition. This appears to
have been the case in the large majority of cartelized industries in
the UK.
Chapter 7 also contains several brief case studies, including the
study of an industry where the free-entry condition seems to have
been violated before the introduction of cartel policy. It is shown
that in this case the theoretical predictions are not confirmed: the
18 Chapter 1

breakdown of collusion led to a fall in profitability and no change in


concentration. This counterexample can be seen as illustrating the
limitations of the present theory, since it suggests that there can be
instances where one of the key assumptions of the theory is violated.
At the same time, it can be seen as a sharp test of the theory’s pre-
dictive power. Thus the counterexample confirms the operation of
the key structural mechanism driving the theoretical predictions
and the econometric results of chapter 7, as well as of much of the
rest of the book—namely, the fact that changes in market structure
are driven by the relationship between firms’ gross profits and sunk
costs.

1.5 Policy Implications

Although this book focuses on positive rather than normative


issues, there are several policy implications of the results presented
here. First, high concentration and mergers may often be a con-
sequence of the intensification of price competition, whether this is
brought about by the successful implementation of cartel policy or
otherwise. This means that there are important economic constraints
on the exercise of merger and antitrust policies. In particular, these
policies cannot be used to impose a market structure so fragmented
that it is not sustainable in a context of intensified competition. To
put this argument slightly differently, some degree of market power
is necessary under conditions of increasing returns to scale, since
firms must be able to cover their sunk costs. In a context of inten-
sified competition, firms will have to be fewer and bigger (on aver-
age) in order to cover their sunk costs, so a stable market structure
cannot be achieved if the necessary restructuring is hindered.
The emphasis on structural mechanisms shaping the evolution of
market concentration—in particular, the emphasis on the relation-
ship between firms’ gross profits and sunk costs—should not, of
course, prevent us from recognizing that concentration is also
influenced by nonstructural factors, and hence that mergers and the
restructuring of industries may go well beyond what is necessary
for market structure to be sustainable. The present study in no way
suggests a lesser role for merger and antitrust policies; rather, it
highlights one of their limitations.
Second, higher concentration need not be associated with higher
profit margins, increased allocative inefficiency, and lower welfare,
An Introductory Overview 19

once free entry is maintained. Therefore, concerns with the level of


market concentration need not take precedence over the need to
ensure that competition remains effective, that is, firms do not en-
gage in collusive practices and no barriers to entry are created. This
strengthens the case for a competition policy that focuses more upon
the monitoring of conduct than on the regulation of market struc-
ture, as has, in fact, always been the approach taken in the UK and
the European Union.
Third, the present study has implications for the much-debated
issue of the relationship between price and nonprice competition. In
particular, it suggests that we need to distinguish between different
types of product-specific investments when evaluating the conse-
quences of price competition for expenditure on product ‘‘quality’’
and for innovation. More specifically, while advertising is likely to
decrease when price competition intensifies, the effect of price com-
petition on the production of innovations seems to be either insig-
nificant or highly ambiguous. Thus the present study offers support
neither for the argument that short-run welfare gains from more
price competition are offset in the longer term by a slower rate of
technological progress nor for the opposite assertion that innovation
is promoted in a context of vigorous price competition. I should also
clarify that I am referring here only to innovations produced by firms.
This book does not examine whether price competition may have an
effect on variable costs, and thus on productivity, through increased
effort by managers and workers, more efficient internal firm organi-
zation, the adoption of better technologies, or otherwise.
Finally, the results of this book imply that the free-entry condition,
which has been popular in recent theoretical work but whose policy
implications are perhaps not fully appreciated, may be a valid prop-
erty of long-run equilibria in the majority of industries, at least as
an approximation. Of course, this does not mean that free entry can
always be relied upon to drive net profits to zero or to maintain
effective competition. For one thing, the presence of asymmetries
between firms implies that only the marginal firm makes zero profit
under free entry. Other firms may well earn higher than normal
profits in the long run, not only because of efficiency differences but
also because of first-mover advantages. Moreover, the focus on long-
run equilibria may be less appropriate in dynamic, innovative indus-
tries than in stable, mature ones. Finally, free entry can be compatible
with restrictive pricing agreements between firms.
20 Chapter 1

What the results of this book suggest is that the level of excess
profits does not depend, in the long run, on firms’ pricing conduct,
because of forces such as entry and exit that push industries toward
the zero-profit equilibrium. It would be misleading to interpret this as
suggesting that competition policy has set itself a relatively easy task
because, by and large, free entry eliminates excess profits in the long
run. Rather, by emphasizing the key role of free entry in restraining
higher than normal profits, this book suggests that, in addition to the
monitoring of firm conduct, the monitoring of entry conditions into
industries is a key priority for competition policy.
2 Cartel Policy and the
Evolution of Competition
in British Industry

2.1 Introduction

Explicit collusive agreements between firms were widespread in


British industry in the mid-1950s: nearly half of the manufacturing
sector was subject to agreements significantly restricting competition.
Some dated from the 1880s and 1890s, many others had been stimu-
lated by government policies for the control of industry during the
two world wars, and still others were the result of the depression of
the interwar years (Swann et al. 1974). The agreements were not en-
forceable at law, but they were not illegal.
The 1956 Restrictive Trade Practices Act required the registration
of most classes of restrictive agreements between firms on goods.
However, there was no outright prohibition of restrictive practices.
Instead, a judicial procedure was adopted for assessing their impact
on the public interest, on a case-by-case basis. As it turned out, the
hard line taken by the newly created Restrictive Practices Court,
especially in its initial judgments, led to the eventual cancellation
of the vast majority of agreements. However, this outcome was dif-
ficult to anticipate before the first decisions of the Court had
been taken. Hence, even though the legislation was introduced in 1956,
it had little effect before the late 1950s or early 1960s. Most collusive
agreements were abandoned between 1959 (the year the first Court
cases were heard) and the mid-1960s.
This chapter begins by describing the key features of the 1956 Act,
as well as some amendments made by legislation introduced in the
1960s, and places the Act within the broader context of British com-
petition policy. It then goes on to discuss the nature, extent, and
effectiveness of collusion in British manufacturing industry in the
1950s. The final part of the chapter provides a detailed account of
22 Chapter 2

the effect of the Act on firm pricing conduct in previously collusive


industries during the 1960s. Drawing on existing evidence from case
studies and descriptive cross-industry surveys, I examine whether
price competition did actually intensify in the large majority of
these industries.

2.2 The Origins of British Cartel Policy

Like competition policy in general, UK cartel policy developed


gradually over several years since the late 1940s. The first piece of
relevant legislation was the 1948 Monopolies and Restrictive Prac-
tices (Inquiry and Control) Act, which embodied a case-by-case
approach to both monopolies and restrictive practices.1 The Mon-
opolies and Restrictive Practices Commission was set up to investi-
gate monopolies and collective restrictive agreements referred to it
by the Board of Trade on a case-by-case basis. The Commission
would normally be asked to determine whether these operated
against the public interest and to recommend appropriate remedies
if required; the government could then take action on the basis of
these recommendations.
There were twenty referrals to the Commission between 1948 and
1950, eighteen of which involved primarily restrictive agreements
between members of industrial trade associations. In the large
majority of these cases, some or all of the restrictions were found to
operate against the public interest. The Commission was generally
more hostile toward market sharing, collective exclusive dealing and
the collective enforcement of resale price maintenance (which often
involved the collective withholding of supplies from retailers selling
below the specified resale prices) than toward price-fixing arrange-
ments. In fact, price-fixing was sometimes not condemned as
being against the public interest, although the Commission usually
recommended in such cases that common prices should be under
some outside supervision or control so as to ensure that they re-
mained reasonable.
The impact of the Commission’s investigations of specific indus-
tries is thought to have been very limited. In several cases an ob-
jectionable agreement was replaced by a more acceptable variety, or

1. Detailed information on the structure of the 1948 Act, the role and the reports of the
Monopolies and Restrictive Practices Commission during the 1950s, and the impact of
these reports is contained in Guenault and Jackson (1974), Rowley (1966), Swann et al.
(1974), and Mercer (1995), among others.
Cartel Policy and Competition in British Industry 23

became subject to government surveillance. Moreover, in some cases


the government did not adopt all of the Commission’s recommen-
dations. Finally, there was no adequate formal machinery to ensure
the implementation of the government’s decisions, and ‘‘follow-up’’
action to secure the continued compliance of the industries involved
was generally not carried out. For all these reasons, none of the in-
dustries investigated by the Commission between 1948 and 1956 ex-
perienced any serious shakeout as a result of changes that took place
following the investigation.
By the early 1950s the slow progress of the Commission was a
cause for some concern. This, together with the fact that the Com-
mission had found certain common types of collective agreements
as being against the public interest in almost every investigated in-
dustry in which these practices occurred, led to a general reference to
the Commission to investigate arrangements involving ‘‘collective
discrimination.’’ These included, among other things, collective ex-
clusive dealing, the collective adoption and enforcement of resale
price maintenance, and the adoption of aggregated rebates and
similar discriminatory practices.2 They did not, however, include
price-fixing or market-sharing agreements. The Commission’s re-
port, published in 1955, is thought to have been very influential in
shaping subsequent developments in cartel legislation. There was, in
fact, a split within the Commission. The majority view was that the
practices concerned should be prohibited, with provisions made for
exceptions on specific grounds. The minority view was that there
should be a call for the registration of agreements and a case-by-
case examination to decide whether each particular agreement oper-
ated against the public interest.
The next stage in the development of legislation on collusive agree-
ments was the introduction of the 1956 Restrictive Trade Practices
Act.

2.3 The 1956 Act

The 1956 Restrictive Trade Practices Act required the registration


of most types of agreements on goods between two or more firms
operating in the UK in which more than one of the parties accepted

2. Aggregated rebates are discounts to distributors on the basis of the total quantity
purchased from association firms. Collective exclusive dealing is the practice whereby
distributors are required to deal only with association firms if they are to be supplied
on best terms or at all.
24 Chapter 2

‘‘restrictions,’’ that is, some limitations on the freedom to make deci-


sions. The Act specified that an agreement was registrable when the
restrictions related to prices to be charged or paid, conditions of sale,
quantities and qualities supplied or acquired and processes used,
persons to or from whom goods could be supplied or acquired,
and areas where goods could be supplied or acquired. Thus the Act
adopted a form-based rather than an effects-based approach to cartel
policy: all agreements of a specified form had to be registered, even
if the restrictions had no significant effect on competition. The defi-
nition of ‘‘agreement’’ included both formal, written undertakings
and informal, verbal, or even implied (as opposed to expressed)
arrangements. It also included cases in which a trade association
made recommendations to its members. The Act also prohibited the
collective enforcement of resale price maintenance. On the other
hand, agreements on services or relating only to exports, most types
of bilateral vertical arrangements, and agreements to exchange
information on prices, quantities, and so on (so-called information
agreements) were exempted from registration. Finally, individual
resale price maintenance was not prohibited; indeed, the Act pro-
vided for its more effective enforcement.3

The Economic and Political Background

Several views have been expressed on the politics of the introduction


of the 1956 legislation. For instance, it has been pointed out that the
1956 Act was partly a response to the new conditions of high growth
and the climate of economic liberalism of the 1950s (i.e., it reflected
the view that the economy would be best served by an increasing
reliance on market forces and a strengthening of competition)
and partly a response to business community dissatisfaction with
the administrative approach through the Monopolies and Restrictive
Practices Commission. In fact, the business community preferred a
judicial review of restrictive practices, which is what the Act estab-
lished. It also seems, however, that the business community was

3. Hunter (1966), Swann et al. (1974), and Wilberforce et al. (1966) describe in detail
the structure and the implementation of the 1956 Act. Subsequent developments in UK
cartel legislation until the early 1970s are discussed in Swann et al. (1974) and also in
the various reports of the Registrar of Restrictive Trading Agreements (RRTA, 1961–
1973). The reports of the Registrar also provide details on the implementation of the
legislation over the period 1956–1972, including brief accounts of cases examined by
the Restrictive Practices Court.
Cartel Policy and Competition in British Industry 25

little prepared for the strict attitude of the Restrictive Practices Court,
which eventually led to the abolition of cartels in Britain.
There is little doubt that the (Conservative) government had to
accommodate a lot of pressure from interested parties, especially
business, during the period of the drafting of the Bill. Hunter (1966,
pp. 85–86) argues that the overall structure of the 1956 Act was a
compromise between business interests, public opinion, and the gov-
ernment’s eagerness to promote economic liberalism. For instance,
he sees the provisions of the Act for more effective enforcement of
individual resale price maintenance and for a dilution of the role
and power of the Monopolies and Restrictive Practices Commission
(which from then on, until the introduction of merger policy in the
UK in the mid-1960s, was called simply Monopolies Commission
and dealt only with monopolies) as major concessions to business
interests. He acknowledges, however, that the structure of the Act
also reflected a willingness not to be dogmatic about restrictive prac-
tices and the recognition of the potential merits of some restrictive
arrangements.
Mercer (1995) provides another interesting, although perhaps more
controversial, account of the origins of the 1956 legislation, empha-
sizing the diversity of interests between different sections of the busi-
ness community, in particular between large, efficient, internationally
oriented firms and smaller, less efficient ones. She argues that the
spread of economic liberalism in the 1950s and the growing discon-
tent of the public and the press with restrictive practices were not the
only factors behind the passing of the 1956 Act, and that the lack of
opposition by larger firms to government plans for the introduc-
tion of cartel legislation was also a key factor. Thus Mercer points
out that some key concessions to industry, such as those mentioned
above, were more likely to benefit larger firms than smaller ones. She
also argues, however, that larger firms did not expect that the Act
would eventually lead to the abolition of a wide range of restrictive
practices. Her argument should not, therefore, be taken to imply that
many or most larger firms actually benefited from the introduction of
cartel legislation.

Key Provisions of the Act

The Act was based on the principle that restrictive agreements were
in general adverse to the public interest unless shown otherwise. A
key feature was the adoption of a judiciary procedure for assessing
26 Chapter 2

the impact of restrictive practices on the public interest. Central to


this procedure was the Restrictive Practices Court, a special judiciary
body consisting of High Court judges and qualified lay members. In
addition, a new officer within the Board of Trade, the Registrar of
Restrictive Trading Agreements, was given responsibility for keep-
ing a register of agreements which was open to public inspection,
referring all but insignificant cases to the Court, and investigating for
any registrable but unregistered agreements. As it turned out, the
process of referring agreements to the Court was gradual and took
several years to complete, partly because the Registrar often tried to
negotiate with the industries concerned modifications to their agree-
ments that would involve dropping all significant restrictions. The
idea was to achieve the purpose of the legislation (as interpreted by
the Registrar) without the need for the industries concerned and the
Registrar to spend time and money on legal proceedings.
The Act prescribed that registered agreements should be aban-
doned unless they were successfully defended in the Court by the
parties or considered by the Registrar as not significantly affecting
competition. Thus there was no outright prohibition of restrictive
practices. An agreement could be allowed to continue if the Court
was convinced that it had benefits for the public, of a kind defined
in the Act, which offset the presumed detriment. In particular, the
Act set out a number of headings, known as ‘‘gateways,’’ by refer-
ence to which an agreement could be defended in the Court. These
related to such things as the protection of the public against injury,
the need to counteract anticompetitive behavior by other firms, the
need to counterbalance monopoly or monopsony power, the avoid-
ance of a serious adverse effect on employment in a specific area,
the avoidance of a serious adverse effect on exports, and, more
generally, the provision of ‘‘specific and substantial’’ benefits to the
public. Thus some of the gateways were sufficiently broad to allow
for almost any agreement to be defended in the Court. Provided
that an agreement passed through one or more of the gateways, it
should then be shown to be in the public interest ‘‘on balance’’ if it
was to be upheld.

The Registration of Agreements

By the end of 1959 some 2,240 agreements had been registered, about
half of which were of nationwide application. Most of them (approx-
Cartel Policy and Competition in British Industry 27

imately 1,800) were between manufacturers, or between manufac-


turers and distributors. Roughly 970 of the national agreements were
thought by the Registrar to be ‘‘important,’’ that is, not dealing with
trivial restrictions or very small industries. Also, 790 of them con-
tained selling price restrictions, and of these, 730 were agreements
made by manufacturers, mostly among themselves rather than with
distributors. In the majority of cases, they were agreements between
members of trade associations. These figures highlight the preva-
lence of pricing restrictions in most collusive arrangements. Other
types of restrictions, such as market sharing or discriminatory prac-
tices, were less common, although this may have been to some ex-
tent the result of redrafting and modification of agreements prior to
registration, as pointed out below.4
An important implication of the procedure set out by the 1956 Act
was that the attitude of the Court could not be known until the first
Court cases had been heard. Thus the large majority of the existing
restrictive agreements were registered rather than being immediately
dropped or secretly continued, although in some cases firms redrafted
their agreements or even removed some of the restrictions in an
attempt to increase the likelihood of a favorable Court decision (see
Swann et al. 1974; Hunter 1966). For instance, a reference to ‘‘agreed’’
prices was sometimes replaced by a reference to ‘‘recommended’’
prices, and market sharing was sometimes dropped. It is interesting
that in those cases where an agreement was modified prior to regis-
tration, firms usually chose to remove those types of restrictions
that had been typically condemned in the Monopolies and Restric-
tive Practices Commission’s reports. Thus collective exclusive deal-
ing was sometimes removed, although price-fixing generally was
not.
The redrafting of agreements indicates that there was a genuine
uncertainty as to how the Act would be implemented and that the
large majority of industries perceived the chances of a favorable
Court decision as being sufficiently high to make registration worth-
while. The ambiguous attitude of the Monopolies and Restrictive
Practices Commission toward price-fixing may have also been a fac-
tor in persuading most industries to register pricing agreements.

4. The actual number of agreements for any given industry has little significance, for
some industries registered only one national agreement covering all or most industry
products, while others registered many agreements, each corresponding to a narrowly
defined product or to a specific region of the country.
28 Chapter 2

One could add that the respectability of price-fixing was well es-
tablished among manufacturers and distributors during the inter-
war years and up to 1956, and, as indicated by various reports of
the Commission during the period 1948–1956, many businessmen
were bewildered by the argument that price-fixing could be thought
of as being against the public interest (see Hunter 1966, pp. 135–
136).
That is not to say, of course, that registration was universal: there
is evidence that a nonnegligible number of agreements were not
registered. Some of these continued secretly, while others were
abandoned. For instance, some of the agreements that had been con-
demned by the Monopolies and Restrictive Practices Commission in
its investigations of particular industries were never registered.
Since these had attracted a great deal of publicity, they were probably
abandoned shortly after the passing of the 1956 Act rather than
being secretly continued. Swann et al. (1974, pp. 153–154) mention
that out of sixty industries with restrictive agreements listed in the
Commission’s report on collective discrimination, eight did not
register their agreements. This is a nonnegligible percentage,
but it has to be borne in mind that in some of these industries the
agreements had comprised only the collective enforcement of resale
price maintenance or collective exclusive dealing, without any fixing
of common prices or market shares. Many of these would normally
not be registered, because the 1956 Act contained an outright pro-
hibition of the collective enforcement of resale price maintenance.5
Once this is taken into account, the incidence of nonregistration of
registrable agreements appears to have been rather low.

Implementation of the Legislation

The adoption of a system whereby collusive agreements were not


illegal per se, and it was the responsibility of a court to examine pos-
sibly complex economic issues in assessing the balance of detriments
and benefits on a case-by-case basis, became a subject of considerable
debate in the UK during the early years of the implementation of
the 1956 Act. However, it did not become a major issue in practice.
The hard line taken by the Restrictive Practices Court in its initial

5. Moreover, collective exclusive dealing had been consistently condemned by the


Commission and was probably dropped by several industries after 1956.
Cartel Policy and Competition in British Industry 29

judgments induced most cartels to voluntarily abandon their agree-


ments rather than incur the costs of a Court case with little hope of
successful defense. Since the first Court cases were heard in 1959,
it was not until that year that industries, on the whole, started aban-
doning their agreements. Of those agreements that came before the
Court and were defended by the parties, only a few were allowed
to continue.
Most of the decisions to allow particular agreements to continue
sparked a great deal of controversy. Some of these cases involved
agreements that were considered by the Court as being beneficial
for exports or for technical cooperation and development, with
the benefits outweighing the detriment of higher prices. In other
cases, the arguments were much more dubious.6 In fact, after an
initial period marked by a series of unfavorable decisions, the
attitude of the Court changed somewhat and a greater proportion
of agreements brought before it were upheld. By that time, how-
ever, the large majority of agreements had been dropped. Most
agreements were formally abandoned between 1959 and the mid-
1960s, although some were canceled as late as the late 1960s. After
the mid-1960s, the attitude of the Court became somewhat tougher
again.7
One of the most significant cases heard by the Court concerned the
agreement between the members of the Yarn Spinners’ Association.
This provided for a minimum selling price for cotton yarn and for
standard terms and conditions of sale. It was the second case to come
before the Court, in early 1959, and the size and historical signifi-
cance of the industry meant that the Court’s decision would be of
considerable importance for industry in general. Moreover, it would
be an indicator of the Court’s likely attitude in future cases, given the
type of arguments put forward by the parties to the agreement and
the general state of the industry. In particular, there was no question
of the cartel firms’ making any excess profits in this declining in-

6. For example, one agreement was allowed to continue partly because it was said to
result in cost savings for buyers because there was no need to ‘‘shop around.’’ Another
was thought to promote product standardization. And still another was allowed to
stand because the Court accepted the argument that the restrictions helped keep prices
below the competitive level by reducing uncertainty, and hence the return on capital
required by firms in the industry.
7. Registration continued during the 1960s and the 1970s. These were mostly agree-
ments of minor significance or agreements that came into operation after the 1956 Act
had come into force.
30 Chapter 2

dustry, and the debate largely focused on the issue of the impact of
the removal of the minimum price scheme on the structure of the
industry, and hence on employment.
The association convinced the Court that the removal of the price
restrictions would be detrimental to employment in specific areas of
the country. At the same time, the Court accepted the Registrar’s
view that the agreement had resulted in prices higher than those
that would have prevailed under competition, and had allowed the
survival of inefficient producers and the maintenance of excess
capacity in the industry, all of which were against the public inter-
est. The Court struck down the agreement on the grounds that its
beneficial effect on employment was outweighed by the detriments
to the public, including the fact that it might have delayed what
seemed to be the necessary restructuring of a declining industry.
It is generally thought that the decision in the Yarn Spinners’ case
was an important factor in inducing a large number of industries to
voluntarily abandon their restrictive practices.
Another important Court case concerned a series of bilateral
agreements between steel firms and a marketing company that
these firms had formed to sell basic slag, a by-product of steel manu-
facture used as a fertilizer. Each of the steel producers appointed
one director to the board of the marketing company and had an inde-
pendent agreement with the company about the terms on which it
supplied it with basic slag, including the obligation to sell the whole
of its output of basic slag to the company. These bilateral agreements
were all identical, and although there was no explicit arrangement
between the steel firms, the Court thought that there was an implied
arrangement between them, which was registrable under the 1956
Act and against the public interest.
The importance of this case lies not in the economic significance of
the product in question, but in the fact that it settled the question
of what constitutes an implied arrangement in UK cartel law. The
Court pointed out that an arrangement between A and B exists if ‘‘(1)
A makes a representation as to his future conduct with the expec-
tation and intention that such conduct on his part will operate as an
inducement to B to act in a particular way, (2) such representation is
communicated to B, who has knowledge that A so expected and in-
tended, and (3) such representation or A’s conduct in fulfilment of
it operates as an inducement, whether among other inducements or
not, to B to act in that particular way.’’
Cartel Policy and Competition in British Industry 31

Information Agreements and Nonregistration

Two other important cases that came before the Court in the mid-
1960s involved arrangements to exchange information on prices and
price changes. Such arrangements were not registrable under the
1956 Act, unless it could be shown that they were being effectively
used as a way to continue a formally abandoned explicit collusive
scheme. In the two cases brought before the Court, there had con-
sistently been information exchanges regarding prices and price
changes in advance of their coming into effect, and the prices of dif-
ferent firms had always been identical. The Court concluded that
these information agreements amounted to the same effect as ex-
plicit price-fixing, and found against the parties to the agreements.
According to Swann et al. (1974), these Court decisions were in-
strumental in inducing several more industries with information
agreements to cancel them in the mid-1960s.
In fact, the extensive use of information agreements was causing
considerable concern among economists and policy makers in the
mid-1960s. Early studies of the impact of the 1956 Act showed that in
many industries explicit price-fixing agreements had been replaced
by informal arrangements to exchange information on prices, price
changes, quantities, and conditions of sale, and that in many cases
these had serious adverse effects on competition. Thus one of the
amendments to the 1956 legislation introduced by the 1968 Restric-
tive Trade Practices Act was a provision for making certain types of
information agreements registrable. It was left to the Board of Trade
to decide which types of information exchanges should be targeted,
however, and only agreements on the exchange of prices and condi-
tions of sale were subsequently called up for registration. By that
time most had already been abandoned.
A second deficiency of the 1956 Act related to the illegal non-
registration of agreements. The Act gave the Registrar of Restrictive
Trading Agreements only moderate powers of discovery. For in-
stance, it did not allow unexpected raids and seizing of company
documents. In addition, there were no significant penalties for non-
registration. In fact, Hunter (1966, pp. 87–88) has argued that the
main deterrent against nonregistration was probably the fear of ad-
verse publicity.
On the other hand, the 1956 Act did provide for penalties in cases
of secret collusion when the firms had been parties to an agreement
32 Chapter 2

that the Registrar had referred to the Court and the Court had struck
down, irrespective of whether the parties had actually defended the
agreement in Court. In some such cases considerable fines were
imposed. This provision was relevant not only for agreements that
had been voluntarily abandoned after having been formally referred
to the Court and condemned, but also for agreements that had been
abandoned before being referred to the Court and condemned. In this
latter case the referral was a precautionary step taken by the Regis-
trar in order to reduce the likelihood of any future anticompetitive
conduct by the firms involved. It is reasonable to assume that these
categories contained the majority of the most serious cases of collu-
sion. Still, a large number of abandoned agreements were not referred
to the Court, so the provision of the 1956 Act regarding penalties
would be irrelevant for firms that had been parties to these (as it
would be for firms that had not been parties to any registered
agreement). The 1968 Act sought to reduce the incentives for non-
registration by allowing injured parties to sue participants to un-
registered agreements for damages. This provision was rarely used,
and it is difficult to know the extent to which it may have deterred
nonregistration.

Resale Price Maintenance

Another important piece of restrictive practices legislation intro-


duced in the 1960s was the 1964 Resale Prices Act. As pointed out
above, although the collective enforcement of resale price mainte-
nance was prohibited after 1956, individual enforcement was allowed.
Under the 1964 Resale Prices Act individual resale price mainte-
nance was prohibited except for industries exempted by order of
the Restrictive Practices Court. The implementation of the legisla-
tion was gradual, and by 1967–1968 resale price maintenance had
been abolished in all industries except for two that obtained exemp-
tions (pharmaceuticals and books). Most of the industries affected by
the 1964 Act were consumer-good, advertising-intensive industries.

An Overall Assessment

The 1956 Act, despite its shortcomings, is generally thought to have


been quite effective in breaking down the previously legal cartels
Cartel Policy and Competition in British Industry 33

in UK manufacturing during the 1960s.8 It may even be the case


that the initial uncertainty about the implementation of the Act con-
tributed to its success. Firms were prompted to register their agree-
ments because they were given the chance (which they perceived as
significant) to have them approved by the Court. Once they had
registered and then formally abandoned the agreements, it may have
been more difficult for them to secretly collude because they were
likely to be monitored by the Registrar and/or their customers, and
also, in many cases, because of the threat of penalties.
In addition, many of the collusive schemes of the 1950s involved
relatively large numbers of firms and were being operated within
industrial trade associations. The legality of the agreements and the
institutional role of trade associations had greatly facilitated the co-
ordination and monitoring of collusion during the 1950s, but this
was no longer possible in the 1960s—which may be part of the rea-
son why many firms perceived registration as their best chance of
being able to continue their collusion. Several more factors that
limited the extent of collusion in the 1960s are discussed in the final
section of this chapter. Before examining the effect of the legislation
on firm conduct in greater detail, however, it is necessary to have a
brief look at the nature of restrictive practices in British industry and
to discuss the available evidence on the effectiveness of the cartels.

2.4 Restrictive Practices in the 1950s

Collusion, although not enforceable in the courts, was not illegal in


the UK in the 1950s.9 The agreements registered under the 1956 Act
revealed for the first time the extent of cartelization in British manu-
facturing. Together with information on unregistered agreements re-
ported in a number of other sources on collusion in British industry,

8. This is also the view taken in the 1979 comprehensive government review of cartel
policy in the UK (SSPCP 1979). Symeonidis (1998) describes subsequent developments
in cartel legislation in the UK, including the legislative attack on restrictive practices in
the commercial services industries in the 1970s, and also provides a general assessment
of the law prior to the 1999 significant restructuring of UK competition policy as a
whole.
9. See Political and Economic Planning (1957) and Kuipers (1950) for the details on the
legal status of the restrictive agreements. The main point is that associations acting in
restraint of trade could not enforce in court agreements between members about con-
ditions on which they transacted business, or agreements to pay penalties. A member
who broke such agreements could not be sued.
34 Chapter 2

they allow us to form a quite accurate picture of the state of compe-


tition across industries at the time the 1956 Act was introduced.
In particular, it is possible to distinguish a group of collusive indus-
tries and a group of competitive industries—as well as a residual
group of industries for which we are uncertain as to their state of
competition. Chapter 3 describes this information in detail. Here I
briefly discuss the nature of the agreements before examining the
available evidence on their effectiveness.

Nature of the Agreements

A large number of agreements involved important horizontal


arrangements, usually between members of trade associations.
A typical agreement of this kind contained agreed or recommended
minimum or fixed producer prices and standard conditions of sale.
Ancillary restrictions, such as collective exclusive dealing or the
maintenance of common resale prices, were often used to strengthen
the arrangement. In some cases, there was no explicit price-fixing,
but the association made recommendations about specified changes
in prices following changes in costs. Some agreements also provided
for market sharing, either through the operation of quota schemes
or by allocating work to parties. In certain cases, especially in some
differentiated product industries, prices were individually set but
there were common maximum discounts to distributors, resale price
maintenance, and sometimes specified conditions of sale or exchange
of information on individual prices and price changes.
In a number of other cases the restrictions were probably of little
significance in terms of their impact on competition. Some agree-
ments related only to ‘‘conditions of sale’’—for instance cash dis-
count terms, delivery charges, formulas for contract price adjustment
when costs changed, nonprice matters, and so on—without any
apparent regulation of prices or trade discounts, even though such
regulation would have been legal under the circumstances. Some
other arrangements provided for the maintenance or even the col-
lective enforcement of individual resale prices, but again neither
individual prices nor trade discounts were regulated. Finally, in
some industries the significance of the restrictions is uncertain, as
in the case of agreements providing for the central notification by
the parties of inquiries received from customers and for some type
Cartel Policy and Competition in British Industry 35

of uniform action, other than price-fixing, with respect to these


inquiries.
The British cartels were primarily price-fixing bodies. Regulation
of nonprice decisions was uncommon. In particular, very few in-
dustries were subject to restrictions regarding capital investment,
media advertising, or R&D expenditure in the 1950s. In fact, media
advertising was not regulated in any advertising-intensive industry,
although restrictions relating to sales promotion (such as gifts and
coupons to consumers, financial inducements to distributors, pub-
licity in distributors’ premises, participation in exhibitions, advertis-
ing in trade catalogs, etc.) were far more common.10 Expenditure
on R&D was noncooperatively determined in all but one or two in-
dustries. There was some degree of technical cooperation, in the form
of exchange of information between firms, in some industries, and
sometimes even some joint R&D. Or patent-owning firms had to
grant licenses to any other firm participating in the arrangement. But
in all but one or two cases these schemes did not amount to anything
close to cooperative determination of R&D expenditure.11
This was certainly true for joint R&D. In nearly all collusive R&D-
intensive industries that performed some joint R&D, such as water-
tube boilers, telephone exchange equipment, or electrical machinery,
the budget of the joint research department was only a small fraction
of the individual firms’ research budgets. For most other collusive
R&D-intensive industries, there is no mention of any R&D coopera-
tion at all. It would also be misleading to put too much emphasis
on the informal exchange of technical information and know-how,
for this often occurred in industries that were not technologically
progressive or related only to basic research and not to product
development. In R&D-intensive industries, on the other hand, the
most important form of technical collaboration was patent pooling,
that is, the obligation of patent-owning firms to grant licenses to
any other firm participating in the arrangement. This was common
in several electrical or electronic engineering industries with price-

10. The only known case of an industry with restrictions on media advertising was
the linoleum industry, a homogeneous-good industry with very low advertising
expenditure.
11. Perhaps the only known case of clear and extensive cooperation in R&D was the
case of the permanent magnet industry, a small industry with modest R&D intensity
and little economic significance.
36 Chapter 2

fixing agreements, including lamps, cables, and telephone exchange


equipment.
However, the effect of patent pooling on R&D expenditure is by
no means clear. While patent pooling may increase R&D spillovers
between firms, its effect on R&D expenditure can be ambiguous and
may depend, for instance, on the exact terms under which licenses
are granted, including the amount of royalties. It may well be the
case that the purpose of the practice was to minimize quality dif-
ferences between firms, and thus facilitate agreement on price (see
chapter 3) rather than to indirectly regulate R&D expenditure. More-
over, there is evidence that patent pooling was practiced in several
electronic engineering industries which were not subject to any pric-
ing agreements (see Burn 1958), so it was not necessarily associated
with collusion. Indeed, its origins in some electronic engineering
industries may be linked to the overall historical context of innova-
tion and international licensing strategies of leading firms in that
sector. In any case, and whatever the origins and effects of patent
pooling, it has to be borne in mind that it was practiced in only a
minority of collusive R&D-intensive industries in the 1950s.

Entry Conditions in Collusive Industries

Was entry restricted in cartelized industries? The evidence from the


agreements registered under the 1956 Act, the reports of the Mon-
opolies and Restrictive Practices Commission (cf. Guenault and
Jackson 1974, Rowley 1966), and the case studies in Swann et al.
(1973) suggests that this was not the general case. Although prac-
tices such as collective exclusive dealing and aggregated rebates
were often used by the cartels as a way of limiting competition from
outside firms, it is not at all clear that these practices also restricted
entry. In most industries the agreements were operated within trade
associations and there were often no significant restrictions on
association membership, so that entry would not be difficult if the
entrant was willing to become a party to the agreement. In some in-
dustries, on the other hand, the existing association members could
reject applications for membership, although they would usually
accommodate any powerful nonmember firm. Obviously, rejecting
an application for membership would be easier in industries where
the association firms had some control over distribution channels,
Cartel Policy and Competition in British Industry 37

usually through agreements with distributors’ associations. Even in


such cases, of course, entry into an industry would not be restricted
unless the barriers to outside competition were fully effective.
The reports of the Monopolies and Restrictive Practices Commis-
sion and the case studies in Swann et al. (1973) also contain infor-
mation on profitability of firms in collusive industries, and this also
supports the view that entry was not restricted in most cases. In par-
ticular, two main conclusions can be drawn from this information.
First, profits were more often thought to be ‘‘reasonable’’ than
‘‘excessive.’’12 Second, there were typically marked variations in
costs and profits across firms, and this often meant that the profit-
ability of the less or least efficient was ‘‘low,’’ and sometimes even
negative in particular lines of production. In fact, it seems that price
setting by the cartels often consisted in a compromise between high-
cost and low-cost firms, with prices set at a level that allowed the
high-cost firms to break even. Of course, it is not clear why cartel
firms should not make excessive profits if they could defend the
profits against entry. The absence of excessive profits in most British
cartels is therefore consistent with a regime of nonrestricted entry.13

Were the Agreements Effective?

The lack of excessive profits does not mean that the agreements were
not effective. The effectiveness of an agreement depended on two
factors: the extent to which the parties conformed to the agreement
or accepted the recommendations,14 and the extent of competition
from outside firms, domestic or foreign. Evidence on these issues
comes from various sources, including the agreements registered
under the 1956 Act; the Monopolies and Restrictive Practices

12. Cartelization in many British industries in the 1950s may have been associated
with high costs and low productivity rather than high profits. See Broadberry and
Crafts (1996) for an econometric analysis of the evolution of labor productivity across
British industries over the period 1954–1963 (which includes the first few years of the
implementation of the 1956 Act) that provides support for this view.
13. I will return to this issue in chapter 7, where I will present strong econometric
evidence consistent with the view that entry was, on the whole, not restricted in car-
telized industries.
14. While some of the agreements provided for sanctions such as a fine or expulsion
from the association in case of nonconformity, most agreements had no such provi-
sions and were therefore entirely voluntary. The provisions regarding fines were not
legally enforceable anyway.
38 Chapter 2

Commission reports; the Political and Economic Planning study of


industrial trade associations, carried out just before the Act was
passed (Political and Economic Planning 1957), and unpublished
background material for this report; and two studies of the impact
of the Act, namely Heath (1961, 1963) and Swann et al. (1973, 1974).
The evidence indicates that the large majority of the agreements were
indeed effective.
In particular, Swann et al. (1974), who conducted case studies
of forty industries that had been subject to collusion in the 1950s,
emphasize that all but one of these agreements had been operated
‘‘honourably’’ before their cancellation. Much of their evidence on
this issue was based on interviews as well as on pricing data over a
number of years prior to cancellation. Swann et al. also found that, in
the large majority of cases, the parties accounted for at least 75% of
the relevant market. Heath (1963) states that 80% of the 159 agree-
ments covered in his questionnaire survey of the short-run impact
of the 1956 Act were reported as being effective at the time of can-
cellation, according to the replies received from firms and trade
associations, while, of the rest, only one in five had been ineffective
for many years. On the other hand, according to the Political and
Economic Planning report on trade associations, some buyers could
purchase at prices lower than the agreed collusive prices, either from
association firms or from outside firms, in about half of the cartel-
ized industries. It was also pointed out in the report, however, that
nonassociation firms were often unable to supply the product in the
quantity or quality required.
There were, in fact, a number of factors that limited the effective-
ness of outside competition in many industries. The cartels tended
to contain most or all of the largest and best-known domestic firms;
the outside firms were often not capable of dealing with large con-
tracts, or manufactured a smaller range of products, or had a lower
reputation for quality. Moreover, practices intended to limit outside
competition—such as aggregated rebates and collective exclusive
dealing—were often used. Finally, competition from imports was
often limited because of tariffs and quantitative controls (see, for in-
stance, Political and Economic Planning 1959, Milward and Brennan
1996), differing technical standards, transport costs, or international
restrictive agreements. Rowley (1966) has even argued that powerful
trade associations were frequently afforded protection on balance-of-
Cartel Policy and Competition in British Industry 39

payments grounds and in order to avoid the development of excess


capacity.15
The available case-study information on collusive prices also sup-
ports the view that the agreements were, in general, effective. Prices
of outside firms were typically lower than the cartel prices, although
sometimes they were identical or only marginally lower. In some
industries, prices were being adjusted according to the extent of
outside competition, for instance, in particular segments of
the market or for particular contracts. More generally, Swann et al.
(1974) found that, although in some collusive industries, prices were
being set on the basis of the costs of the more efficient producers,
in most cases they were being determined either on the basis of a
weighted average of all costs or largely arbitrarily, since no system
of relating prices to costs existed or accounting procedures were de-
ficient. The view that prices had been kept, in general, above com-
petitive levels is further strengthened by the evolution of prices and
profits after the cancellation of the agreements. It is precisely to this
latter issue and other aspects of the impact of the 1956 legislation
on competition that I will now turn.

15. The limited extent of foreign competition across a range of UK industries in the
1950s is reflected in the low level of import penetration. A rough calculation at the
four-digit industry level suggests that, in more than 80% of manufacturing industries,
imports accounted for less than 10% of the total value of sales in the domestic market.
Even in industries with significant imports—for instance, 50% or more of total sales in
the domestic market—the high level of imports does not necessarily reflect competi-
tive pressure. In fact, in many industries with high import penetration, the imported
goods were largely complementary to domestic products within the relevant four-
digit industry (for example: preserved fruits and vegetables, oils and fats, nonferrous
metals, timber). In other cases the industries concerned were subject to quantitative
controls or other forms of regulation (thus imports in bacon, iron ore, and other prod-
ucts were very high, but they were a residual source of supply).
Of course, this should not be taken to imply that all UK manufacturing industries
were relatively free from foreign competition. Sectors facing a relatively low degree of
protection included most food and drink industries, some chemicals, basic metals,
clothing and footwear, wood products, publishing, leather and most textile semi-
manufactures, and building materials. Note, however, that in several of these indus-
tries, foreign competition was limited due to transport costs or international collusive
agreements. Sectors with a relatively high degree of protection included the engineer-
ing industries, instruments, vehicles, finished metal goods, some chemicals, paper and
paper products, furniture, pottery and glass, most finished textile goods, rubber prod-
ucts, and various other finished manufactures. See Political and Economic Planning
(1959), Morgan and Martin (1975), Kitchin (1976), and Milward and Brennan (1996) for
details on these issues.
40 Chapter 2

2.5 The Impact of the 1956 Legislation on Firm Conduct

The two necessary conditions for a significant effect of the 1956


Act on competition in British industry are (1) that the price-fixing
agreements had generally been effective and the parties to them had
accounted for a substantial fraction of the relevant markets, and (2)
that the explicit agreements were not replaced—at least not in the
long run—by informal arrangements having the same effect on com-
petition. In the previous section I discussed the available evidence on
the former condition. Here I examine the evidence on the latter.
Some information on the impact of the 1956 legislation on compe-
tition is contained in the Registrar’s reports (RRTA 1961–1973). In
the second report, covering the period 1960–1961, the emphasis was
on the ‘‘forces of inertia,’’ but in the third, covering the period 1961–
1963, the emphasis was different and it was noted that in some
industries there was now ‘‘keen competition’’ between the previously
colluding firms. Sometimes the breakdown of collusion had been
triggered by a specific event, such as a fall in demand or the entry of
a new competitor. The fourth report, covering the period 1963–1966,
qualified this optimistic view in light of the persistence of infor-
mation agreements across a range of industries. This concern was
replaced in subsequent reports by a more general concern about
cases of secret collusion. On the whole, however, the Registrar’s
reports paint a picture of slow but steady emergence of competition
in previously collusive industries.

The Short-Run Effect

There have also been two more systematic studies of the impact of
the 1956 Act on competition. Heath (1961, 1963) conducted a ques-
tionnaire survey of the short-run consequences of the Act, using a
sample of 159 canceled price agreements. The survey was conducted
in 1960, only about a year after industries started formally abandon-
ing explicit price-fixing. In only about a third of the cases did the
respondents think that competition had intensified. This can be ex-
plained by short-term inertia or reluctance to abandon long-standing
practices. In fact, the results of the survey seem to strengthen the
view that the majority of industries had not anticipated the hard line
taken by the Restrictive Practices Court toward cartels, and were not
prepared for a more competitive environment in the short run. Heath
Cartel Policy and Competition in British Industry 41

also noted that many canceled arrangements had been replaced by


information agreements.
A more extensive study of the impact of the Act was carried out
by Swann, O’Brien, Maunder, and Howe in the early 1970s (Swann
et al. 1973, 1974). They conducted case studies of forty industries,
including thirty-four with canceled agreements and six with agree-
ments upheld by the Court, and they examined both the short-
run and the long-run effects of the legislation. The industries in the
sample were chosen to reflect, as the authors pointed out, a wide
variety of industrial situations, including producer-good as well
as consumer-good industries, capital-intensive as well as labor-
intensive industries, and so on.
Regarding the short run, the authors found that price competition
increased in about half of the thirty-four industries with canceled
agreements. This took the form of a fall in list prices or an increase
in discounts to distributors. In particular, price falls of 15% to 25%
occurred in six cases. In the other cases the price fall was smaller
and/or the increase in competition took the form of an increase in
trade discounts. On the other hand, information agreements re-
placed the former arrangements in at least half of the industries
with formally canceled agreements in the sample and were usually,
but not always, successful in restricting competition.

The Long-Run Effect

With respect to the long run, Swann et al. found that in many indus-
tries with information agreements, price competition emerged
after these were abandoned or changed from prenotification into
postnotification arrangements in the mid-1960s, following the two
decisions of the Restrictive Practices Court on information agree-
ments discussed earlier in this chapter. As a result, price wars oc-
curred in a number of previously collusive industries in the second
half of the 1960s, and the final blow came with the provisions of
the 1968 Restrictive Trade Practices Act regarding information agree-
ments. Thus in most industries with information agreements, com-
petition emerged about a decade after the introduction of the 1956
legislation.
While the overall picture was one of a gradual strengthening of
competition in British manufacturing industries, Swann et al. also
noted that in the longer term there were instances of secret collusion
42 Chapter 2

or unregistered information agreements in some industries and par-


allel pricing (i.e., similar and roughly simultaneous price changes)
in others. It is necessary to discuss this point in some detail, for it
might seem to suggest that in many industries with abandoned
agreements, the intensity of price competition may have not in-
creased significantly in the long run. This is not the case. For one
thing, the number of cases of detected explicit collusion was small.
On the other hand, parallel pricing was quite common: the Mon-
opolies Commission report on parallel pricing (Monopolies Com-
mission 1973) listed approximately sixty industries in which the
witnesses thought that parallel pricing occurred. Should this be seen
as evidence for the existence of effective (explicit or tacit) collusion in
many of these industries?16 The answer is probably not, and it is
based both on case-study evidence and on a more general assessment
of changes in the competitive environment facing British firms in the
second half of the 1960s.
In particular, evidence from Swann et al. (1973, 1974), Maunder
(1972), and various reports of the Monopolies Commission and the
National Board of Prices and Incomes on specific industries suggests
that in many cases there was competition in discounts and with
respect to sales by tender to large buyers, even though list prices
tended to move in parallel; and that, in several industries that were
reported as being subject to parallel pricing, increases in list prices
were not always being followed. Second, the operation of prices and
incomes policy since the mid-1960s, while facilitating parallel pricing
in some industries, must have prevented excessive increases in profit
margins. Third, the progressive opening of the British economy dur-
ing the 1960s and 1970s must have significantly reduced the scope
for effective collusion by domestic firms. And finally, the extent of
collusion must have been limited by the operation of the cartel policy
itself, including the right of buyers to complain to the competition
authorities when they suspected collusion, the powers of the Regis-
trar to investigate, and the provision for penalties to be imposed on
colluding firms which had been parties to agreements that the Court
had struck down. Swann et al. also mention the lack of moral com-

16. The distinction between collusion and parallel pricing implied here is not just one
of form, but also one of effect. Thus effective collusion between a given number of firms
is assumed to always increase profit margins above ‘‘competitive’’ levels, while this is
not necessarily the case for parallel pricing.
Cartel Policy and Competition in British Industry 43

mitment compared to the pre-1956 agreements as an additional fac-


tor limiting collusion in the 1960s.17

2.6 Concluding Remarks

The 1956 Restrictive Trade Practices Act was quite effective in dras-
tically reducing the extent of cartelization in British manufacturing
between the 1950s and the 1970s. In particular, the large majority of
collusive industries registered their explicit restrictive agreements,
and then had to abandon them. The process was slow, and it took
more than a decade for the effects of the law to be fully realized.
Admittedly, there is some uncertainty about the impact of the legis-
lation on competition in each particular industry. Some agreements
may have not been effective at the time they were formally canceled,
and others may have been replaced, even in the longer term, by
secret or tacit arrangements. More generally, the ‘‘degree of collu-
sion’’ can vary, depending on the type of restrictions, the extent of
outside competition, the balance of interests within the cartel, and so
on—and so can the ‘‘intensity of competition’’ in the absence of
any explicit collusion. Is it then legitimate to reduce this continuum
of states to just two competition regimes and interpret the increase
in price competition following the cancellation of agreements as a
change in regime?
A cautious approach would be to say that for an industry that
abandoned an explicit restrictive agreement, there is a high proba-
bility that there was, at some point after the cancellation of the
agreement, a change of competition regime induced by the legisla-
tion, while this probability is zero for an industry not affected by
the Act. The empirical analysis carried out in this book, based on a
comparison of the evolution of the two groups of industries, would
still be valid under this interpretation. To the extent that some agree-
ments were ineffective in the 1950s and others were replaced by

17. Another finding of Swann et al. was that prices or price-cost margins in some of
the industries where competition broke out were restored or partly restored in the long
run. This does not imply, however, a weakening of competition. As described in sev-
eral of the case studies contained in Swann et al. (1973), this often occurred either after
the number of firms in the industry had decreased, in which case it is perfectly con-
sistent with competition remaining effective (see chapter 7 of this book), or because
prices had fallen to very low levels during a price war or due to a temporary fall in
demand.
44 Chapter 2

undetected secret or tacit collusion in the 1960s and the 1970s, the
analysis would, if anything, only somewhat understate the mag-
nitude of the effect of price competition on the endogenous vari-
ables of interest, namely, market structure, nonprice conduct, and
profitability.
I think, however, that there is no reason to emphasize this more
cautious interpretation. The survey and case-study evidence re-
viewed in this chapter suggests that the large majority of industries
that dropped their explicit collusive agreements as a result of the
1956 legislation did experience, sooner or later, an intensification of
price competition. Furthermore, the effect was in many cases sig-
nificant. Hence it is, on the whole, valid to think of this evolution
as a change of competition regime induced by an exogenous insti-
tutional change. Consequently, a comparison between those indus-
tries affected by the 1956 Act and those not affected should provide
an accurate assessment of the effects of competition.
3 The Competition Data

3.1 Introduction

The analysis of the effects of competition in this book is based on a


comparison of those industries with a change of competition regime
following the 1956 Restrictive Trade Practices Act and those without
a change in regime. This chapter begins with a detailed description
of the available data on competition and collusion across British
manufacturing industries in the 1950s and after. The data allow us
to distinguish a group of collusive industries, the large majority of
which subsequently experienced a change of competition regime;
a group of industries without explicit collusive agreements, which
therefore experienced no change in regime; and a residual group of
industries for which we are uncertain as to their state of competition.
The criteria for constructing these groups will be explained in detail,
and the issue of potential measurement error or selection bias due to
unknown cases of collusion will also be discussed extensively. I will
argue that this latter issue is probably not a significant problem for
this chapter, and is certainly not a serious difficulty for the analysis
carried out in the rest of the book. I will also provide some descrip-
tive statistics on the extent of cartelization across sectors. A detailed
survey of restrictive agreements across industries is contained in
appendix A.
The final part of the chapter contains an analysis of the structural
industry characteristics associated with the incidence of explicit
collusion in the UK in the 1950s. Although the analysis relates to
explicit collusion, it has to be borne in mind that restrictive agree-
ments in the UK were not enforceable at law, so the issues of coor-
dination and enforcement of agreements were just as relevant as
they would be in a context where collusion is tacit. Thus this chapter
46 Chapter 3

aims to provide more general insights regarding the factors facilitat-


ing or hindering collusion. At the same time, it prepares the way for
the study of the effects of the breakdown of collusion in the rest of
the book by examining the similarities and the differences in the ini-
tial conditions in the two groups of industries, those with explicit
restrictive agreements and those without. A key result is that the two
groups were quite similar with respect to initial market structure, but
rather different regarding capital intensity, advertising effectiveness,
and technological opportunity. It seems, then, that the incidence of
collusive pricing in British industries in the 1950s was largely deter-
mined by exogenous variables, some of which are unobservable in-
dustry characteristics, rather than by endogenous variables such as
concentration.

3.2 Data Sources on Competition in British Industry

The most comprehensive source of data on competition in British


industry in the 1950s and the 1960s is the Register of Restrictive
Trading Agreements created under the 1956 Act. This register has
always been open to public inspection; in recent years, it has been
kept at the Office of Fair Trading in London.
For each agreement in the register, information is available on the
products subject to restrictions, the types of restrictions, the names of
the parties, the geographical coverage, and the time period during
which the agreement was in force after the introduction of the 1956
Act. The amount of detail on the content of any given agreement in
the register can vary greatly, depending on what was furnished by
the parties and, especially, on how elaborate any given agreement
was in the first place. For example, in several cases there are elabo-
rate price lists, covering a large number of different products and
specifications, with details on terms and conditions of sale, and
possibly also lists of buyers participating in an ancillary scheme of
collective exclusive dealing or entitled to carry the products or to
receive specified discounts. In many other cases there are again price
lists, but they are less elaborate. And in some cases, there is little
more than a page containing brief evidence of an agreement to ob-
serve certain restrictions, or possibly a simple recommendation by a
trade association to its members to raise their prices by a specified
amount in order to cover the latest cost increase. In several files, ad-
The Competition Data 47

ditional information is given in the constitution and by-laws of the


relevant trade association, or in the minutes of meetings between the
parties.
For those agreements that were brought before the Restrictive
Practices Court and were defended by the parties, there is also sup-
plementary, and much more detailed, information in the reports of
the Court. These contain detailed accounts of the Court proceedings,
and they are often very informative, not only about the exact way in
which the agreements were operated but also about the key features
of the industries in question, thus allowing for a better understand-
ing of the purpose and significance of particular restrictions. Need-
less to say, agreements brought before the Court and defended by
the parties can safely be assumed to have been effective in the 1950s.
As for the rest, there is no guarantee that all registered agreements
were indeed fully effective at the time, but the case-study evidence
reviewed in chapter 2 suggested that this is a reasonable assumption
to make, and that any resulting measurement error should be small;
this issue is further discussed below.
The register of agreements and the Court reports contain infor-
mation on explicit collusion not only in the 1950s but also over the
entire period examined in this book. Thus there is information on
the dates when the agreements were formally abandoned, on Court
decisions to allow explicit collusion to continue in certain industries,
and on agreements discovered by the Registrar and his staff during
the 1960s and the 1970s.
There is also evidence that some agreements were not registered,
either because they were abandoned shortly after the Act was intro-
duced or, in a few cases, because they secretly continued. A number
of sources can be used to identify products subject to unregistered
agreements. These sources sometimes provide additional informa-
tion on registered agreements as well, either because some agree-
ments were modified prior to registration or because some sources
contain much more detailed information on the industries concerned
than the registered agreements. Some of the sources are not as reli-
able as others, or provide less detailed information.
The reports of the Monopolies and Restrictive Practices Commis-
sion before 1956 and those of the Monopolies Commission after that
date are entirely reliable and provide significant detail on the in-
dustries concerned. It is reasonable to assume that those collusive
48 Chapter 3

industries that were investigated before 1956 and did not register
their agreements (or registered modified versions of them) must have
abandoned them (or abandoned the unregistered restrictions) soon
after the introduction of the legislation. It is indeed unlikely that in
these much-publicized cases explicit collusion would have secretly
continued. The situation with industries investigated after 1956 is
slightly more complicated. These were usually investigated because
of the presence of a dominant firm or a dominant group of firms. In
the majority of these cases, either there is no mention of any restric-
tive arrangements before or after 1956 or there is a reference to some
previously registered agreements. In a few cases, on the other hand,
an industry is reported as having had an agreement in the 1950s
which either was abandoned, and therefore never registered under
the 1956 Act, or was modified before registration. Finally, there are a
few cases where the Monopolies Commission uncovered an existing
secret collusive agreement during its investigation; information is
then provided on the origins of the arrangement.
For the purposes of my classification, all restrictive agreements
mentioned in the Commission’s reports as having been effective prior
to 1956 (or after that date) were treated in exactly the same way as
registered agreements—or, more precisely, as registered agreements
defended in the Court, since there was no question in those cases of
the restrictions not being effective.
Other reliable sources on the state of competition across manu-
facturing industries in the 1950s include the 1955 Monopolies and
Restrictive Practices Commission report on collective discrimination
(MRPC 1955), and several industrial case studies undertaken in the
mid-1950s as part of an investigation into the structure of British in-
dustry and contained in Burn (1958). Various other publications with
occasional references to competition in particular industries in the
1950s and the 1960s are mentioned in appendix A. There were only a
few agreements reported in these sources and not in any others, and
they were treated in the same way as registered agreements for the
purposes of my classification.
Finally, two very important but not fully reliable sources are the
Political and Economic Planning study of industrial trade associa-
tions, undertaken just before the 1956 Act was passed, and the Board
of Trade annual reports on the operation of the 1948 Monopolies and
Restrictive Practices Act, covering the years 1949 to 1956 (Board of
Trade 1949–1952, 1953–1956). These sources provide information on
The Competition Data 49

industries alleged to be collusive. Most of these industries registered


agreements, but some did not.1
More specifically, the Board of Trade annual reports contain lists
of products allegedly subject to restrictive practices during the period
1949–1956. This information is not very detailed as to the types of
restrictions or the coverage of the alleged agreements. More im-
portant, it is based on complaints made by buyers to the Board of
Trade, and buyers may wrongly deduce the existence of a price-
fixing agreement from parallel pricing, or even from mere price
uniformity over a certain time period. It is not very clear to what ex-
tent the complaints received by the Board of Trade were subject to
some preliminary evaluation before being used to compile the lists
of suspect industries, and it is certainly emphasized in the reports
that inclusion of a product does not necessarily imply that a restric-
tive agreement actually exists. On the other hand, it may well be
the case that some of the industries that are on the lists and did not
register any agreements were in fact collusive. Thus, there is some
uncertainty regarding the state of competition for products in the
Board of Trade annual reports but not mentioned as being the sub-
ject of restrictive agreements in any of the more reliable sources. For
this reason, these products were classified as having an ambiguous
state of competition in the 1950s. This also implies some ambiguity
as to the effect of the 1956 Act in these cases.
Information on restrictive agreements in particular industries is
also provided by the Political and Economic Planning (P.E.P.) (1957)
study of industrial trade associations. Most of this information is
found in the unpublished background material for this study, for the
published report provides mostly aggregate data at the sector level.
This background material has been deposited in the archives collec-
tion of the British Library of Political and Economic Science. Again,
much of the information is not very detailed, and comes mostly from
buyers (public and private firms, local authorities, etc.). For the same
reasons as in the case of the Board of Trade annual reports, this
information must be treated with some caution. Hence most indus-
tries that were alleged to be collusive in the P.E.P. study, and are
not mentioned as collusive in any of the more reliable sources, were

1. The reverse is also true: several industries that registered agreements do not appear
as collusive in the Political and Economic Planning survey or the Board of Trade an-
nual reports. On the whole, these two sources probably underestimate rather than
overestimate the extent of cartelization in British industry in the 1950s.
50 Chapter 3

treated as ambiguous for the purposes of my classification. On the


other hand, a fair amount of information in the P.E.P. survey comes
directly from firms that were parties to agreements. This information
can be treated as reliable and used to identify a few industries that
were certainly collusive but failed to register their agreements.
In summary, the above sources allow us to distinguish three groups
of products: a group with explicit restrictive agreements, a group
without explicit restrictive agreements, and a group with uncertain
state of competition in the 1950s. The first of these groups contains
all products for which there are registered agreements or which are
reported as being subject to collusion in the Monopolies Commis-
sion’s reports or any of the other fully reliable secondary sources.
The second group contains products not mentioned in any of the
various sources. And the third group contains products that do not
appear in the register or in any of the other fully reliable sources, but
may have been subject to explicit or tacit collusion according to the
Board of Trade annual reports or the P.E.P. survey.
This classification is based on certain assumptions as to the relia-
bility of the various data sources which seem very reasonable. There
is also, however, a more fundamental assumption implicit in this
classification: that the existence of a known (i.e., documented) ex-
plicit price-fixing agreement is a good overall indicator of collusive
conduct in British manufacturing industry in the 1950s. One might
raise a number of objections to this, and it is now necessary to dis-
cuss these in some detail.

The Reliability of the Competition Data: Tacit versus Explicit


Collusion

One possible objection is that the data on British cartels relate to ex-
plicit collusion, but not to tacit collusion. Could it then be the case
that some of the industries classified as noncollusive actually prac-
ticed tacit collusion? One cannot, of course, rule out the possibility
that firms in an industry colluded tacitly in the absence of any ex-
plicit arrangement. However, it is difficult to understand why col-
luding firms would not want to enter into an explicit arrangement,
given that explicit collusion was legal and widespread in the 1950s.
Furthermore, the 1956 Act required the registration of informal and
even ‘‘implied’’ understandings as well as formal agreements, and
this seems to cover cases of tacit collusion. So tacitly colluding firms
The Competition Data 51

would be in little doubt as to their obligation to register. Of course, it


is still possible that they might have chosen not to do so—but this is
a different issue, to be discussed presently. As far as tacit collusion is
concerned, I would argue that, for all the above reasons and given
that collusive arrangements of all kinds were not enforceable in the
courts, the distinction between tacit and explicit collusion is not very
important in the present context.

The Reliability of the Competition Data: The Effectiveness of


Agreements

Another possible objection is that some of the agreements may have


not been effective at the time they were registered, so classifying these
industries as collusive introduces measurement error. There is an
even stronger version of this argument: that some of the industries
that registered explicit agreements may have done so because collu-
sion was difficult to sustain otherwise. Therefore we may be assign-
ing to the collusive group industries where collusion was actually
rather difficult to sustain. This would clearly be worse than mea-
surement error, for it would essentially amount to selection bias. Are
the above arguments valid? For one thing, there is little doubt that
some of the agreements registered under the 1956 Act may not have
been effective. The important question is whether these cases are a
large fraction of the total number of registered agreements, and, if
so, whether this is likely to result in selection bias. I think that the
answer to these questions is in the negative, and I will offer several
arguments to support this claim.
First, the case-study evidence discussed in chapter 2 strongly sup-
ports the view that the large majority of the agreements had been
effective. Recall that only one out of the forty agreements examined
in Swann et al. (1974) was thought by the authors to have been in-
effective. Second, the Register of Restrictive Trading Agreements is
not the only source of information on collusion in British industry.
Several industries were investigated by the Monopolies and Re-
strictive Practices Commission during the 1950s, and several more
defended their agreements before the Restrictive Practices Court.
The available information, which in these cases is quite detailed,
leaves no doubt as to the effectiveness of these agreements, which
are a significant part of the total number. Third, all the sources of
information on the effects of the 1956 Act, including the Registrar’s
52 Chapter 3

reports, emphasize that competition was slow to emerge in many


industries, and that this was often due to the fact that information
agreements replaced the former price-fixing arrangements in the short
run. This is not consistent with the view that these arrangements
were not effective before 1956.
Fourth, it should be emphasized that a weak agreement could
not expect to gain much from a favorable Court decision, because
it would still not be enforceable at law. So it is not at all clear why
industries with weak agreements would have a strong incentive to
register. On the contrary, one might argue that it was industries
with strong agreements that had the strongest incentive to register
and try to maintain collusion, because of the potentially large cost
of a cartel breakdown. And, finally, the evidence from Lydall (1958)
and Board of Trade (1944, 1946) discussed below also suggests that
there is no serious measurement error or selection bias in the con-
struction of the collusive group of industries in the present study.

The Reliability of the Competition Data: The Issue of Nonregistration

The above discussion suggests that the assumption that the existence
of an explicit price-fixing agreement is a good overall indicator of
collusive conduct is not an unreasonable one in the present con-
text. There is, however, one final difficulty, which is probably more
serious, for it relates to the issue of nonregistration. Again, there is a
weaker and a stronger version of the argument. The weaker version
is that nonregistration of agreements, if widespread, would lead to
serious measurement error. The stronger version is that certain types
of industries may have had a stronger incentive to avoid registration
than others. Then the failure to take unknown cases of collusion into
account could lead to sample selection bias.
There are two possible reasons for nonregistration: firms may sus-
pend an agreement, so that there is nothing to register, or they may
switch to secret or tacit collusion. Take, first, the former case. A rea-
sonable conjecture in this case is that very weak agreements would
be more likely to be dropped immediately than stronger ones. Even
if that were true, we would not be losing much by failing to identify
such cases, since we are primarily interested in identifying effective
agreements, not ineffective ones. But it is not even clear why the
decision to immediately cancel an agreement rather than register it
should have occurred in certain types of industries more than in
The Competition Data 53

others, and might therefore be associated with industry character-


istics related to the sustainability of collusion: there is not much to
be lost by registering an agreement, even a weak one, when the
alternative is cancellation. In fact, many of the agreements that were
not registered were those that had been condemned by the Monop-
olies and Restrictive Practices Commission. Clearly, these were not
weak agreements, but the parties may have thought that they had
practically no chance of success in the Court and wished to avoid
further adverse publicity. That leaves us with the second reason for
nonregistration mentioned above. One might argue, for instance,
that industries where tacit or secret collusion would be easier to
sustain or less easily detected after 1956 had less of an incentive to
register. Failure to identify such cases could cause sample selection
bias.
An important thing to note with respect to the issue of non-
registration is the historical context of the introduction of the 1956
Act. As pointed out in chapter 2, there was a great deal of uncer-
tainty about the way the legislation would be implemented. It seems
that industries genuinely thought that they had a good chance of
success in the Court, as evidenced by the fact that many agreements
were redrafted prior to registration and several others were defended
in the Court despite the first few unfavorable Court decisions. The
ambiguity of the attitude of the Monopolies and Restrictive Prac-
tices Commission toward certain types of restrictions, including
price-fixing, must have contributed to creating this perception
among firms. Recall also that a comparison of the register with a
list of industries subject to restrictive agreements published in the
Commission’s report on collective discrimination in chapter 2 re-
vealed a low incidence of nonregistration of registrable agreements.
To these arguments one could add that the 1956 Act gave the
Registrar powers of investigation. Being an officer at the Board of
Trade, the Registrar would certainly have access to all the complaints
made throughout the 1950s by buyers claiming the existence of re-
strictive agreements in particular industries. Thus it would be diffi-
cult for many industries to collude secretly and go unnoticed for a
long time. For this and all the other reasons mentioned above, it
seems safe to conclude that nonregistration was not a widespread
phenomenon.
Moreover, and most important, in this book I do not rely solely on
information about registered agreements. As pointed out above,
54 Chapter 3

several sources were examined to obtain information on agreements


that were not registered. Most of these contain long lists of industries
that either were certainly collusive or were alleged to be collusive.
The Board of Trade, in particular, was encouraging buyers to come
up with information and complaints, which it then used to compile
its lists of allegedly collusive industries throughout the period 1949–
1956. Although these sources are far from perfect, it would really be
surprising if there were a significant number of cases that escaped all
of them.
There is also some evidence from a questionnaire survey of com-
petition in UK manufacturing in the 1950s (Lydall 1958) which does
not support the view that there may be significant selection bias in
the data. The study by Lydall used a sample of 876 manufacturing
firms from all sectors. It did not specifically discuss collusive agree-
ments, but some of the information provided suggests that firms that
perceived their condition as being characterized by ‘‘no strong com-
petition’’ were primarily in industries which had a high incidence of
explicit collusion, according to my classification, while firms that
thought that they were facing ‘‘strong competition’’ were chiefly in
industries without many agreements. Thus the three sectors with the
greatest proportion of firms saying that they faced strong competi-
tion were chemicals, food and drink, and textiles. As shown later in
this chapter (table 3.1), in these three sectors the number of compet-
itive industries, according to the classification adopted in this book,
was much larger than the number of collusive ones.2 In addition,
Lydall mentions engineering, other metal products, and paper and
printing as the three sectors with the highest proportion of firms tak-
ing part in price-fixing. Again, table 3.1 below reveals that in these
three sectors the number of competitive industries, according to my
classification, was far smaller than the number of collusive ones.
The final piece of evidence comes from two surveys of cartels and
restrictive practices carried out in the mid-1940s by the Board of
Trade, one dealing with international cartels in which British firms
had some involvement (Board of Trade 1944), the other dealing with
internal cartels (Board of Trade 1946). Of course, these surveys were
carried out more than a decade before the introduction of the 1956
Act and describe, for the most part, the state of competition across

2. It was in fact much larger in chemicals and food and drink, and only slightly larger
in textiles. However, the textiles sector is mentioned by Lydall as one with a high
proportion of firms acknowledging the existence of price leadership.
The Competition Data 55

manufacturing industries during the 1930s and the early 1940s. Given,
however, that most of the British cartels of the 1950s had been active
at the time of the Board of Trade surveys (see Swann et al. 1974), a
comparison of the information contained in the surveys with my
own data on collusion in the 1950s would certainly be interesting.
The two surveys differ somewhat in scope: the one on international
cartels was meant to be comprehensive, while the one on internal
cartels was conceived as the first part of a wider investigation, and
was therefore somewhat less comprehensive, although the industries
or sectors chosen spanned the whole spectrum of manufacturing
industries, covering capital-good industries as well as consumer-
good industries.
Of the two surveys, the one on internal cartels is probably the
more useful for my present purposes. The number of industries
reported in this survey as being subject to price-fixing and related
restrictions is equal to almost a third of the total number of indus-
tries classified as collusive in my data set (see table 3.1). In other
words, the coverage of the Board of Trade survey was quite large.
In spite of this large coverage, a careful reading of the report re-
vealed virtually no cases of industries reported as being subject to
restrictive agreements in the late 1930s or early 1940s which were not
mentioned as being collusive in any of my primary data sources
for this book. Furthermore, nearly all of these industries registered
agreements under the 1956 Act.3

An Overall Assessment

It seems reasonable to conclude, on the basis of the above discussion,


that the issue of potential nonregistration of agreements does not
cause any significant bias or measurement error in the present data.
Admittedly, it is not possible to back up this claim with anything
more than the indirect evidence discussed above. Let me therefore

3. The Board of Trade survey of international cartels is less relevant for my present
purposes because in many cases only one British firm—usually a dominant firm in the
relevant UK industry—participated in the international agreement. Moreover, inter-
national cartels often broke down during or after the Second World War. Nevertheless,
the report also mentions several agreements in which more than one British firm was a
party. Most of these are also reported in my data sources for the 1950s, although a few
are not. It is impossible to know whether these agreements were still effective in 1956.
Even if they were, this would change my classification of industries in only three cases.
56 Chapter 3

clarify the precise implications of accepting this claim as a main-


tained assumption for the analysis carried out in the rest of this book.
Except for section 3.4 of the present chapter, which examines the
cross section of industries in the 1950s to identify how the incidence
of collusion was associated with various structural industry charac-
teristics, this book focuses on the effects of the breakdown of cartels.
The key distinction therefore is between industries with a change in
competition regime and industries without such a change. Seen from
this angle, the issue of potential nonregistration followed by collu-
sion after 1956 is not very important. Any industry that practiced
(explicit or tacit) collusion successfully, did not register any agree-
ment, is not mentioned in any of my sources as being collusive, and
continued to collude after 1956 would be classified as an industry
without a change in competition regime. This is indeed the correct
classification: it does not really matter whether the lack of change
was due to continuing collusion or to continuing competition.
In other words, all the results reported in chapters 4–7 of this book
are not sensitive to the existence of cases of unknown and continuing
collusion. The only measurement error in these results can be caused
by agreements that were ineffective in the 1950s or agreements that
were registered and then replaced by undetected secret or tacit col-
lusion. This can only understate the estimated effects of price compe-
tition on the endogenous variables of interest. Bearing this in mind,
let me now turn to a detailed description of the competition data.

3.3 The Construction of the Data Set

The approach to modeling the competition effect in this book in-


volves distinguishing between those industries with a change of
competition regime following the 1956 Act and those without a
change in regime. The first step in this procedure was to assess the
reliability of the various sources of information on competition in
British industry in the 1950s; this was discussed in the previous sec-
tion. The next step was to classify all industries according to their
state of competition in the 1950s on the basis of three criteria: (1) the
reliability of the sources of information on any particular agreement;
(2) the types of restrictions; and (3) the proportion of an industry’s
total sales covered by products subject to agreements and, for each
product, the fraction of the UK market covered by cartel firms.
The Competition Data 57

Types of Restrictions

The following types of restrictions were classified as significant


with respect to their effect on competition: (1) agreed or recom-
mended minimum or fixed producer prices—possibly complemented
by maximum or fixed trade discounts, terms and conditions of sale
or contract, resale price maintenance, collective exclusive dealing,
and so on; (2) agreed or recommended price changes; (3) agreed
maximum or fixed discounts to distributors or users combined
with resale price maintenance and terms and conditions of sale—
and possibly also prenotification of price or product changes; and (4)
quotas, market sharing, or allocation of work to parties.
The following were classified as not significant: (1) terms and
conditions of sale or contract, relating to price and/or nonprice
matters, without any regulation of prices or trade discounts; and (2)
resale price maintenance, individual or collective. Finally, one type of
restriction—the reporting of inquiries received from customers
leading to some kind of uniform action on terms and conditions,
but generally not on prices—was considered as having an uncertain
effect on competition.
This classification covers the large majority of the schemes oper-
ated by British firms at the time the 1956 legislation was passed. The
remaining few cases were intermediate or special, and were usually
classified as uncertain (see appendix A for details).
It is necessary to discuss certain aspects of this classification in
some detail. Consider, first, agreements that contained maximum or
fixed discounts to distributors or users, as well as the maintenance of
individually set resale prices, but did not regulate the level of indi-
vidual prices. Most of these schemes were operated in differentiated-
product industries, where the fixing of common prices may have not
been practicable. They typically contained elaborate discount struc-
tures, which were often supported by ancillary restrictions relat-
ing to sales promotion. In some cases, they had succeeded earlier
price-fixing agreements that had been modified prior to registration.
Finally, case-study evidence suggests that the cancellation of these
agreements led to significant increases in trade discounts in some
industries. What all this indicates is that this type of restriction was
usually a substitute for, or a weaker form of, price-fixing; it was
therefore classified as significant.
58 Chapter 3

Second, consider the practice of resale price maintenance in the


absence of any agreement between manufacturers specifying either
what these maintained prices would be or at least a common set of
trade discounts from these prices—other than, possibly, the fixing of
minimum discounts in order to protect distributors’ margins. Some-
times there was an agreement between firms to maintain their resale
prices, and sometimes there was no such agreement but resale price
maintenance was practiced on an individual basis by most firms in
an industry. Also, the enforcement of resale price maintenance could
be, until 1956, either individual or collective (for instance, by means
of a stop list).4 These distinctions are not very important in the
present context, however, for there are good reasons to believe
that, in the absence of any horizontal arrangement on prices, resale price
maintenance did not significantly affect competition between
manufacturers in the industries where it was practiced until it was
abolished following the 1964 Resale Prices Act.
The literature on resale price maintenance suggests that the main
reason why this practice may reduce competition among manufac-
turers is that it may help sustain collusion by facilitating the detec-
tion of cheating and by minimizing fluctuations of manufacturers’
market shares caused by competition among retailers (see Overstreet
1983, Yamey 1966a). This argument is valid in general, but it is prob-
ably not very relevant in a context where horizontal price-fixing
agreements are widespread and are not illegal. In many industries
with agreed common or minimum prices, resale price maintenance
may indeed have served as an ancillary restriction to facilitate the
monitoring of cheating. But consider an industry without any agree-
ment between manufacturers specifying either what the prices would
be or at least a common set of trade discounts. If firms had actually
intended to restrict competition among themselves, they would also
have explicitly fixed prices or at least maximum trade discounts. The
failure to do that in the circumstances of the 1950s can only imply
that resale price maintenance was being used in that industry either
to protect distributors’ margins or as part of an individual pricing
strategy.

4. Under an agreement among manufacturers to collectively enforce resale price main-


tenance, a distributor who cut the price of the product of one manufacturer was liable to
punishment by all manufacturers. Such punishment often involved a refusal to supply
the distributor concerned.
The Competition Data 59

An interesting observation in this respect is that in industries


where collective resale price maintenance was practiced without any
horizontal pricing arrangement among manufacturers, the agreement
to maintain retail prices was usually an agreement between a manu-
facturers’ association and a distributors’ association.5 This is consis-
tent with Yamey’s (1966b) view attributing the origins of resale price
maintenance in many industries to the pressure exerted on manu-
facturers by retailers’ associations.6 Moreover, there is evidence (e.g.,
Board of Trade 1949, Pickering 1966) of unrestricted and even in-
tense competition between manufacturers in several industries
where resale price maintenance was practiced, either individually or
collectively.
For all these reasons, I have chosen to classify resale price mainte-
nance per se as a nonsignificant restriction in the present context. A
possible objection to this is that this practice may in some cases have
facilitated tacit or secret collusion in previously cartelized industries
after 1956. However, it would then, at most, cause only a temporary
delay in the emergence of price competition, comparable to that
attributed to information agreements. This is because resale price
maintenance gradually weakened or broke down in several indus-
tries after 1956 and was finally abandoned in almost all industries in
the mid-1960s (roughly the same time that information agreements
were dropped).
Third, consider those arrangements that related solely to agreed
or recommended terms and conditions of sale or contract. This class
covers a wide range of restrictions on price and nonprice matters.
Examples are the recommendation not to give trade discounts; for-
mulas for contract price adjustment to cover costs when delivery
does not take place until some time after the contract is made and
costs have changed in the meantime; restrictions on the amount of
credit given to customers; agreed or recommended cash discount

5. This is in sharp contrast with agreements that contained maximum or fixed dis-
counts to distributors or users in addition to the maintenance of individually set resale
prices. These agreements were often among manufacturers only. As mentioned above,
I have classified such schemes as significant restrictions on competition.
6. Admittedly, Yamey’s view has not been without its critics, partly because of evidence
that manufacturers in many industries opposed the abolition of resale price mainte-
nance at the time of the introduction of the 1964 Resale Prices Act. Still, this opposition
was attributed by most of these critics to the presumed efficiency gains from resale
price maintenance rather than to any potential effect of this practice on competition
between manufacturers (see Mercer 1998).
60 Chapter 3

terms, delivery or packing charges, terms of guarantee, arrange-


ments for returns of unsold goods, and so on.
It is difficult to say to what extent such restrictions may reduce
competition rather than simply help maintain orderly and stable
trading conditions. However, in the absence of any arrangement con-
cerning prices, such practices should not affect the state of competi-
tion to any significant degree, especially in a context where collusion
is not illegal. As in the case of resale price maintenance, one could
argue that if firms had actually intended to restrict competition, they
would also have explicitly fixed prices. This interpretation is consis-
tent with the fact that several agreements of this kind were allowed
to continue, although firms were often required to modify certain
restrictions before their agreements were cleared.7
Finally, consider those agreements that involved the reporting
of inquiries received from customers and uniform action on terms
and conditions with respect to these inquiries, but did not provide
for price-fixing or the allocation of work. The sharing of information
about inquiries from prospective purchasers was not uncommon in
industries where the costs of tendering for a job were significant, as
was the case in many engineering industries. Often the sharing of
information was part of a price-fixing or market-sharing agreement
between firms. But in some cases the arrangement did not provide
for price-fixing or the allocation of work following the circulation of
information about received inquiries. Furthermore, evidence from
the Monopolies and Restrictive Practices Commission reports and
the P.E.P. survey suggests that such arrangements were not simply
the product of redrafting of agreements prior to registration.
There are two reasons why I classified this type of restriction per
se as uncertain. In many cases these agreements also provided for
the payment by the successful tenderer of a sum of money to be used
to compensate other tenderers for the costs of tendering. In some
industries the amounts involved were small, while in others they
were on a larger scale and may have successfully raised the level of

7. The view taken by UK competition authorities, as expressed in the 1980 report of


the Director General of Fair Trading (DGFT 1981), was that the recommendation by
trade associations of standard terms and conditions of sale or contract—such as terms
of guarantee, provisions as to contractual liability for loss or damage, formulas for
contract price adjustment, and so on—may often be beneficial to all parties involved
and will have no significant effect on competition, provided that the recommended
terms and conditions are ‘‘fair’’ to both buyers and sellers, and are clearly stated as
being optional rather than compulsory.
The Competition Data 61

prices. However, information on the level of the tendering fee is not


always available. More important, these arrangements may have oc-
casionally led to informal understandings about prices or the alloca-
tion of work, even if they did not provide for this. Overall, then, it
is difficult to assess the extent to which these agreements restricted
competition without detailed information on how they were actually
operated in each particular case.8 In the absence of such information,
the safest approach was to classify this type of restriction as having
an uncertain effect on competition.9

Classifying the Industries According to Their State of Competition


in the 1950s

Once the various types of restrictions were classified as significant,


nonsignificant, or uncertain, the next step was to assign the products
that were subject to agreements to the various industry categories
used. It was often the case that certain products within a particular
industry were subject to significant restrictions while others were
not. In this book several different data sets are being used, mainly
because the industry definitions used in the published statistics for

8. The view expressed here is not inconsistent with the view taken by the Monopolies
and Restrictive Practices Commission in its report on electrical machinery (MRPC
1957). Although most sections of the electrical machinery industry had explicit price-
fixing agreements in the 1950s, a few had only agreements to report inquiries received
from prospective purchasers. The Commission concluded that this practice might be
expected to operate against the public interest because of the danger that under-
standings about prices or allocation of work would follow. Thus the Commission es-
sentially recognized the uncertain effect of such arrangements. In the context of the
heavily cartelized electrical machinery industry, however, the Commission was prob-
ably justified in stressing the potential for collusion. In other cases, the assessment
would have been less straightforward.
9. Note that I have not specifically discussed collective exclusive dealing and other
forms of collective discrimination, such as aggregated rebates. The reason is that these
were typically used as ancillary restrictions to strengthen a pricing agreement. How-
ever, in very few industries collective exclusive dealing was practiced even though the
manufacturers did not collude on prices or trade discounts. These cases present
something of a paradox: If collective exclusive dealing was being used as an entry
barrier, why didn’t these firms also attempt to restrict price competition among them-
selves? Since some of these cases are discussed extensively in reports of the Monopolies
and Restrictive Practices Commission, there can be no question of insufficient infor-
mation about these industries. A possible explanation is that these were instances of
bilateral collective exclusive dealing, the primary purpose of which was to protect the
interests of distributors rather than those of manufacturers. Fortunately, the industries
concerned are very few, and either they are not included in the statistical data sources
used for this book or they have been classified as ambiguous anyway.
62 Chapter 3

the various endogenous variables are often very difficult to match.


Thus different industry categories must be used when analyzing the
competition effect on concentration, advertising intensity, innova-
tions, and profitability. Obviously, the use of different industry
categories requires creating multiple competition data sets. How-
ever, the criteria for constructing these data sets were always the
same, and they were obviously independent of the industry defini-
tions used in each case.
In particular, an industry was classified as collusive in the 1950s
if the products subject to significant restrictions accounted for more
than 50% of total industry sales revenue. It was classified as com-
petitive if the products subject to significant or uncertain restrictions
accounted for less than 10% of industry sales revenue. And it was
classified as ambiguous in all remaining cases.10 Information about
the particular products within each industry and the sales revenue
of each product was taken from the Census of Production. For
agreements of nationwide application, it was assumed that the
parties accounted for a substantial fraction of the relevant market
(an assumption that is in line with the case-study evidence). For
important regional agreements, an estimate of the fraction of indus-
try sales subject to restrictions was made on the basis of the avail-
able information on the geographical coverage of the agreements
and the regional structure of production in the industries concerned.
Some discussion regarding the use of the 10% and 50% cutoff
points is in order. In particular, there are two main questions: First,
why choose these particular cutoff points? Second, why use any cutoff
points?
Clearly, the choice of these particular cutoff points is arbitrary.
However, it should be emphasized that most industries classified
as competitive were free from any restrictive agreements. Similarly,
most industries classified as collusive had agreements covering all
industry products. I have used the 10% cutoff point because in some
cases secondary industry products were subject to restrictive agree-
ments, although core industry products were not. It was clear, how-
ever, that these secondary products did not significantly affect the
structure and performance of the industries concerned. Also, I have
used the 50% cutoff point because in some cases most core industry

10. Products mentioned as being subject to collusion or potential collusion in one of


the less reliable data sources, but not in any of the more reliable ones, were treated as
products with uncertain restrictions for the purposes of this classification.
The Competition Data 63

products were subject to price-fixing, although some were not. Surely,


one would expect a significant effect of the 1956 Act in these indus-
tries. In any case, I have also experimented with alternative cutoff
points, in particular 20% instead of 10%, on the one hand, and 40% or
70% instead of 50%, on the other. These variations have changed the
samples used very little, and in some cases not at all. None of the
results reported in this book were significantly affected by variations
in the chosen cutoff points.11
One might ask why it is necessary to use any cutoff points at all,
since an alternative would presumably be to use the information on
the exact fraction of sales revenue covered by products with agree-
ments in each industry, on the assumption that the higher this frac-
tion, the more collusive the industry (and thus, all else being equal,
the stronger the effect of the Act on competition). However, this
approach is impractical, or even misleading, for several reasons.
First, the fraction of sales revenue covered by products subject to
agreements in each industry is sometimes difficult to specify other
than approximately. This can occur because the information is not
sufficiently detailed to allow for this fraction to be estimated very
precisely, even though it is sufficient for classifying the industry
according to the cutoff points I have used. An additional difficulty
is that the exact fraction of sales revenue covered by products with
agreements may vary across years. This is not a significant problem
when using cutoff points, but it is more serious when trying to con-
struct a continuous measure of competition.
Second, there is no one-to-one link between the fraction of sales
revenue covered by products with agreements in each industry and
the state of competition across industries in the 1950s. The link be-
tween the two is blurred by a variety of factors, including differences
across industries regarding the types of restrictions, the extent of

11. This also implies that the classification of marginal cases makes little difference to
the results. There were, in fact, very few marginal cases (i.e., cases where the fraction
of industry sales revenue subject to restrictions was either very close to 10% or 50%, or
was higher than one of these cutoff points in one year and lower in another year).
These cases were classified on the basis of two criteria. The first was the average or
typical value of this fraction over all relevant years. The second was an assessment of
which industry products were mainly driving the industry-level values of the endog-
enous variables of interest in each particular case. For instance, in the case of concen-
tration, I compared the average sales revenue of firms producing regulated products
with the average sales revenue of firms producing nonregulated products, the as-
sumption being that the market shares of the larger firms within an industry will
largely drive the concentration ratio of the industry as a whole.
64 Chapter 3

outside competition, and the extent to which competition in export


markets was regulated (see below). For many of these factors, sys-
tematic information is not available. Thus, even though it is clear
that an industry with 90% of sales revenue covered by products
subject to agreements can be seen as collusive, while one with only
5–10% of sales revenue covered by regulated products can be re-
garded as competitive, it is not at all easy to compare, say, 90% with
75%, or 40% with 25%.
Third, it is often the case that some products within an industry
were subject to significant restrictions in the 1950s, while other prod-
ucts were subject to uncertain restrictions. It is not clear how to deal
with such cases if one wants to construct a continuous measure of
competition based on the fraction of industry sales revenue under
agreements.
Fourth, it is sometimes the case that an industry consists of sub-
divisions with very different market structures, and although some
of these subdivisions were collusive in the 1950s, most were not. The
fraction of sales revenue covered by products with agreements is
then between 10% and 50%, but it is very difficult to draw any con-
clusions about the state of competition in the industry as a whole
and its effect on other variables. This crucially depends on which
industry subdivision is mainly driving the industry-level measures
of concentration, profitability, and so on. The use of cutoff points
has the advantage of treating such industries for what they really are
(ambiguous), rather than trying to fit them into a continuum of states
of competition.
Finally, let me again emphasize that the use of cutoff points is
actually far less ad hoc than it may seem at first sight, because the
majority of industries classified as competitive were essentially free
from restrictive agreements and most industries classified as collu-
sive had agreements covering practically all industry products.

Exclusion of Ambiguous Industries

Once all manufacturing industries were classified as collusive, com-


petitive, or ambiguous in the 1950s, according to the criteria set out
above, those industries with ambiguous state of competition were
excluded from the samples used in this book. The purpose of this
was to be able to compare two clearly defined groups, one collusive
and one competitive. Treating ambiguous industries as intermediate
The Competition Data 65

was not really appropriate, given the heterogeneity of this group,


which includes cases with uncertain restrictions, cases for which
there is insufficient information, and cases where the regulated
products cover between 10% and 50% of total industry sales. The
heterogeneity of this group of industries also implies that their
exclusion should not lead to any significant selection bias in the re-
sults reported in subsequent chapters, since the selection is not done
on the basis of any of the endogenous variables of interest.12

The Incidence of Collusion in the 1960s and the 1970s

Having classified all industries as collusive or competitive in the


1950s, the final step in constructing the competition data for exam-
ining the effect of the 1956 Act in this book was to take into account
the available information about cases of collusion in the 1960s and
the 1970s. There were no such cases among industries without ex-
plicit collusive agreements in the 1950s. On the other hand, although
the large majority of industries that were collusive in the 1950s did
experience a change of competition regime, there were two kinds of
exceptions.
First, as pointed out in chapter 2, a small number of collusive
agreements were allowed to continue. In most of these cases, the
agreements were still in force in the mid-1970s, so I assigned the
industries concerned to the group that was not affected by the 1956
legislation.13 Second, some illegal restrictive arrangements were

12. The potential selection bias due to exclusion of ambiguous industries may be more
of a concern for the results on the determinants of collusion presented in the final sec-
tion of the present chapter. However, as pointed out below, these results were found to
be robust to alternative classifications of the industries according to their competitive
status in the 1950s.
13. Except when (1) the fraction of the industry covered by the continuing agreement
was less than 50% or (2) the fraction of the industry covered by the continuing agree-
ment was more than 50% but another fraction of the same industry, between 10% and
50%, was affected by the Act. Such cases were classified as ambiguous. I should also
emphasize that the decisions by the Restrictive Practices Court to allow certain agree-
ments to continue do not appear to have been determined by any specific character-
istics of the industries involved (see the discussion in chapter 2). Nor was the evolution
of these industries similar during the 1960s. Take, for instance, the two collusive
industries included in my sample for chapter 4 and assigned to the group that was not
affected by the 1956 legislation: cement and glazed tiles. In the cement industry the
five-firm concentration ratio increased from 0.86 in 1958 to only 0.89 in 1968, while the
price-cost margin increased from 0.25 to 0.29 during the same period. On the other hand,
in the glazed tile industry the five-firm concentration ratio increased from 0.65 in 1958
to 0.94 in 1968, and the price-cost margin fell from 0.19 to 0.18 during the same period.
66 Chapter 3

made during the 1960s in previously cartelized industries and were


still in force in the early or mid-1970s (before being discovered and
referred to the Court). Most of the industries in this group were
excluded from the samples used in chapters 4–7 for various rea-
sons—for instance, because the fraction of industry sales revenue
covered by products subject to collusion was lower than 50%, or be-
cause the agreement was very unstable or the restrictions were not
very significant. One has to acknowledge that these industries may
not be a random sample, but they are very few anyway. Details on
these and other special cases are provided in appendix A.

Export Agreements

A small complication that has arisen with the above method of clas-
sifying the industries relates to the operation of export agreements.
As mentioned in chapter 2, export agreements were not registrable
under the 1956 Act. Thus information on whether any given collu-
sive industry had a restrictive agreement for the export market in
addition to the one for the home market is often not available. Fur-
thermore, little is known about whether and when any particular
export agreement was abandoned. Now if exports in a collusive in-
dustry were never subject to restrictions, or if they were subject to
restrictions but remained so after 1958, then the impact of the 1956
legislation on competition in the industry would be smaller than
what we would measure it to be under the assumption of no differ-
ence between the home market and the export market. Only in the
case where the restrictions originally applied to both markets and
were subsequently dropped in both markets would we be measuring
things correctly.
It could be argued that this is indeed what happened in many,
probably most, cases. It is clear from the available information that
a large number of agreements covered exports as well as domestic
sales. Also, once collusion had broken down in the domestic market,
it may have been difficult for firms to maintain it in the export market,
especially in the context of intensifying international competition
of the 1960s and the 1970s. To this argument one could add that
the whole issue is of no importance for industries with insignificant
exports.
Still, the issue should not be dismissed so easily. In a few cases,
we know that the export market was not subject to collusion, and
we can take this into account when calculating the fraction of in-
The Competition Data 67

dustry sales subject to significant restrictions in the 1950s. More


often, however, either we know that exports were regulated but we
do not know for how long, or we have no information at all. In such
cases I have assumed that the export market was affected by the 1956
Act in the same way that the domestic market was. Note that the
effect of the potential misclassification of industries caused by this
assumption can only be an overestimation of the impact of the 1956
Act on competition. In other words, some industries that should re-
ally be excluded from the sample as ambiguous may be classified in
the group of industries with a change of competition regime in my
data set, and some that should be really classified in the control
group may be excluded from my data set as ambiguous. Thus the
results obtained in chapters 4–7 of the book might understate, if
anything, the effect of price competition on the endogenous variables
of interest.
However, even this possibility is very remote. To check the maxi-
mum potential misclassification of industries due to export agree-
ments, I estimated the approximate ratio of exports to total sales in
1958 for all the industries with available concentration ratios for that
year, and I reclassified the industries, using the 10% and 50% cutoff
points, on the assumption that the export market was not affected
by the 1956 Act. I found only two industries whose classification
changed, which means that the whole issue can be ignored for all
practical purposes, at least as long as industries are classified on the
basis of cutoff points such as the ones I have used.

Summary of the Classification Criteria

This concludes my discussion of the construction of the competi-


tion data. To summarize, an industry in my data was assigned to the
group of industries that experienced a change of competition regime
if it had been subject to significant collusive agreements in the 1950s
covering at least 50% of total sales revenue and these agreements
were subsequently abandoned. It was assigned to the control group
if less than 10% of the industry was affected by the legislation. In-
termediate cases, defined as ambiguous, were excluded.

A Comparison with the Broadberry and Crafts Classification

Broadberry and Crafts (1996, 2000), who have carried out a separate
survey of restrictive agreements in British industry in the 1950s, have
68 Chapter 3

used somewhat different criteria than the ones that I use here to clas-
sify industries as collusive, competitive, or ambiguous in the 1950s.
As a result, their categorization of industries is different from mine
in several cases, although in almost all of these the difference is by
one class (i.e., collusive versus ambiguous, or ambiguous versus
competitive). Broadberry and Crafts use the three-digit level of aggre-
gation, while I typically use more disaggregated industry categories.
A side effect of this is that the divergence between our respective
classifications is smaller than it would have been if I had used
the three-digit level of aggregation as well. More specifically, there
are three main differences between their classification criteria and
mine.
First, Broadberry and Crafts rely heavily on registered agreements,
and make relatively little use of other sources on collusion in British
industry. For instance, they do not mention the Monopolies and Re-
strictive Practices Commission reports or the Board of Trade annual
reports. They do report some information from the Political and Eco-
nomic Planning survey, however, although it is not clear how they
are using this information when classifying the industries. A com-
parison of their lists of industries with mine suggests that this factor
accounts for part, although not most, of the divergence between our
respective classifications.
Second, their two principal criteria for assessing the significance of
registered agreements are (1) the existence of a detailed price list
with nationwide application, and (2) whether the agreement was re-
ferred to the Restrictive Practices Court by the Registrar or at least
specifically mentioned in the Registrar’s reports as being a pricing
agreement. These criteria are somewhat different from the ones that I
use to classify the significance of particular types of restrictions (see
above). For instance, I do not regard price-fixing agreements that
contain a detailed price list as being necessarily more significant or
more effective than ones which do not. Clearly, the choice of criteria
for assessing the significance of particular agreements is to some ex-
tent subjective. As it turns out, however, the difference between the
Broadberry and Crafts approach and my approach in this respect
does not result in any major discrepancies between our respective
classifications of industries, partly because most of the ambiguous
cases end up in the intermediate category both in their categorization
and in mine. Nevertheless, this factor adds to the divergence be-
tween our respective classifications.
The Competition Data 69

Third, it is not clear whether or how Broadberry and Crafts are


using the proportion of an industry’s total sales covered by products
subject to agreements as a criterion in their classification. A compar-
ison of their lists of industries with mine suggests that they have
classified as competitive some industries with very low incidence of
collusion, just as I have done, but at the same time they have classified
as collusive several industries with agreements covering consider-
ably less than 50% of total industry sales. My approach is different,
and this also accounts for a good part of the divergence between our
respective classifications.

The Incidence of Collusion across Sectors

Table 3.1 summarizes the available information on the occurrence of


collusion across manufacturing sectors in the 1950s. The industry
definitions used in this table are taken from the official concentration
statistics for 1958, as reported in summary table 5 of the Report on the
Census of Production, 1963 (London: H.M.S.O., 1967), and correspond

Table 3.1
Competition and collusion across manufacturing sectors in the 1950s
No. of industries classified as
Collusive Ambiguous Competitive
Mining and quarrying 4 0 0
Food, drink, and tobacco 8 8 14
Chemicals and allied products 1 4 8
Basic metal manufacture 12 1 2
Mechanical engineering 6 10 8
Instrument engineering 0 3 2
Electrical engineering 12 2 5
Shipbuilding and vehicles 0 4 2
Other metal products 9 4 2
Textiles 8 9 9
Leather, leather goods, and fur 0 0 4
Clothing and footwear 0 1 11
Bricks, pottery, glass, cement, etc. 8 4 1
Timber, furniture, etc. 1 1 4
Paper, printing, and publishing 5 2 3
Other manufacturing industries 2 3 5
Total 76 56 80
70 Chapter 3

to Census four-digit industries.14 The 212 industries included in


table 3.1 cover most of the manufacturing sector, although certain
sectors, such as chemicals or instruments, may be somewhat under-
represented, and others, such as basic metals or textiles, may be
slightly overrepresented. Since this sample will also be used in the
next section to examine the links between the incidence of collusion
and other variables, industries without available concentration data
for 1958 are not included. However, there is no reason to believe that
this causes any significant bias with respect to comparisons across
sectors.
On the whole, table 3.1 provides a fairly accurate picture of car-
telization in British manufacturing in the 1950s, distinguishing be-
tween collusive industries, competitive industries, and industries
with ambiguous state of competition. Table 3.2, which uses the same
sample of industries as table 3.1, summarizes the evolution of com-
petition between 1958 and 1975 across sectors, distinguishing be-
tween industries with a change in competition regime and industries
not affected by the 1956 Act (and omitting any ambiguous cases).15 A
much more detailed account of the evolution of competition across
all manufacturing industries can be found in appendix A.
As can be seen from these tables, collusive agreements were much
more widespread in some sectors than in others. In particular, sec-
tors with a high or very high incidence of collusion (measured as the
fraction of collusive industries in the total number of industries in
any given sector) include mining and quarrying, basic metals, other
metal products, building materials, and electrical engineering. Sec-
tors with a low or very low incidence of collusion include leather
goods, clothing and footwear, and instruments. In the case of chem-
icals, the relatively low incidence of collusion shown in table 3.1 is to
some extent a feature of the particular sample used, since a number

14. In constructing this sample, I have excluded four industries with significant gov-
ernment participation or intervention in the 1950s (sugar, ordnance and small arms,
aircraft, locomotives), two industries that hardly existed prior to the mid-1950s (syn-
thetic rubber, tufted carpets), and one where imports were very much higher than
domestic production and were subject to regulation (iron ore).
15. In comparing the two tables, recall that some collusive industries experienced no
change in regime, and for others the evolution was ambiguous because of agreements
with uncertain effect on competition that were made in the 1960s. Another reason for
the discrepancies in the classification of industries across the two tables is that for table
3.1 the 10% and 50% cutoff points are applied to the period 1951–1958, while for table 3.2
they are applied to the period after 1958. The correlations reported in section 3.4 below
are based on table 3.1, but this of course is not the case for the results of later chapters.
The Competition Data 71

Table 3.2
The evolution of competition across manufacturing sectors after 1958
No. of industries with
Change in No change in
competition competition
regime between regime between
1958 and 1975 1958 and 1975
Mining and quarrying 1 0
Food, drink, and tobacco 8 15
Chemicals and allied products 1 8
Basic metal manufacture 12 2
Mechanical engineering 5 8
Instrument engineering 0 2
Electrical engineering 11 5
Shipbuilding and vehicles 1 2
Other metal products 8 3
Textiles 7 9
Leather, leather goods, and fur 0 4
Clothing and footwear 0 11
Bricks, pottery, glass, cement, etc. 5 3
Timber, furniture, etc. 1 4
Paper, printing, and publishing 3 3
Other manufacturing industries 2 5
Total 65 84

of chemical industries with restrictive agreements are not included in


the sample. However, the figures also reflect the absence of agree-
ments in most of the advertising-intensive, consumer-good indus-
tries in the chemical sector.
Of the other sectors, paper, printing, and publishing has an above-
average incidence of collusion; textiles and mechanical engineering
are intermediate; and food and drink, shipbuilding and vehicles,
wood products, and other manufacturing have a below-average inci-
dence of collusion. These statistics reveal little about the structural
industry characteristics associated with the incidence of collusion in
the 1950s, however. This is the subject of the next section.

3.4 The Determinants of Collusion

The data on the British cartels of the 1950s provide a very rare op-
portunity to examine the impact of several structural industry char-
72 Chapter 3

acteristics on the incidence of collusive pricing, using information


which is unusually comprehensive and, as argued above, relatively
free from selection bias. An extensive and detailed analysis of the
determinants of collusion in British manufacturing industry has been
carried out elsewhere (Symeonidis 1999b). Here I will present only
some basic statistics that highlight and summarize most of the results
obtained, with a focus on results that are particularly relevant as a
background for the rest of this book.
In particular, I will discuss how the incidence of explicit collusion
is related to a small number of basic structural industry character-
istics, namely, capital intensity or the level of sunk entry costs, con-
centration, and the scope for advertising or R&D. As it turns out, the
results reported below are robust to several extensions of the present
approach, including the use of econometric analysis, the adoption
of a two-stage estimation procedure in order to deal with potential
endogeneity issues, the introduction of several more control vari-
ables, the use of alternative proxies for some of the variables of in-
terest, and the use of modified criteria to classify industries according
to their competitive status in order to check for any potential selec-
tion bias (see Symeonidis 1999b).

Theoretical Background

Firms that collude have generally two problems to solve. They must
coordinate on a price or a set of prices. And they must ensure that
collusion is enforceable, that is, that none of the parties has an in-
centive to cheat. Thus a common distinction in the literature on the
factors facilitating or hindering collusion is between coordination
and enforcement of collusive agreements.
Product homogeneity, high industry concentration, the absence
of significant cost or size differences between firms, institutional
features such as the existence of an industry trade association, and
sociocultural characteristics such as ‘‘homogeneity of values’’ and
social ties among senior managers are often cited as factors facilitat-
ing coordination on a collusive price or set of prices (see, e.g., Scherer
and Ross 1990). In addition, coordination must be easier under a
permissive climate of competition policy or when collusive agree-
ments are not illegal. While these ideas are intuitively appealing,
they have proved difficult to model within a formal game-theoretic
framework.
The Competition Data 73

On the other hand, the issue of enforcement of collusive agree-


ments has led to a substantial formal literature on cartel stability.
These models can be used to analyze both tacit and explicit collusion
when firms cannot legally enforce agreements. They are all formal-
izations of the basic idea that firms may be able to sustain collusion
by threatening to retaliate at a future time in the event of a price cut
today. This literature has produced a number of unambiguous pre-
dictions regarding the factors that promote cartel stability. These in-
clude low uncertainty about demand, conditions that facilitate the
monitoring of rival firms, and short retaliation lags (see Tirole 1988,
Martin 2002). Another common finding is that high concentration
facilitates collusion, although Brock and Scheinkman (1985) have
argued that the link may be nonmonotonic in the presence of ca-
pacity constraints. Finally, this literature predicts that collusion is
hindered by differences in quality or ‘‘perceived quality’’ between
firms in vertically differentiated markets (Häckner 1994, Symeonidis
1999a), and will therefore occur less frequently in industries with
high advertising effectiveness, and possibly also in industries with
high technological opportunity, than in exogenous sunk cost
industries.
One problem in much of the theoretical literature on collusion is
that market structure is taken as given and entry is not explicitly
modeled.16 It may be argued, however, that any factors that limit
the extent of entry (and, more generally, the volatility of market
shares) facilitate collusion, because coordination and monitoring are
easier when the identity of the principal competitors does not change
much over time, and also because the gain from collusion will be less
quickly or less easily eliminated by entry. Since capital intensity im-
plies significant sunk costs of entry, this argument can provide a
theoretical rationale for a positive effect of capital intensity and the
level of sunk entry costs on the likelihood of collusion.
To summarize, the theoretical literature suggests a number of fac-
tors that are expected to facilitate or hinder collusion. Some of them
are difficult to observe or cannot be tested in empirical work using
cross-section data. Of those that are in principle testable, capital
intensity or the level of sunk entry costs, and possibly also concen-
tration, are thought to facilitate collusion, while demand uncertainty

16. See Fershtman and Pakes (2000) for a recent theoretical approach to collusion that
allows for entry and exit.
74 Chapter 3

and the scope for product differentiation, including vertical differen-


tiation through advertising or R&D, are expected to hinder collusion.

Previous Empirical Evidence on the Determinants of Collusion

Despite the rich theoretical literature on the subject, empirical studies


of the determinants of cartel formation and sustainability using di-
rect evidence, rather than relying on profitability indices to infer the
possible operation of collusive arrangements, are relatively rare. An
early study of collusive pricing in the UK in the 1950s by Phillips
(1972) did not obtain clear results, partly because of data limitations.
Other studies, using US data, have each produced clearer results, but
the overall picture is rather mixed. Thus, while the findings by Hay
and Kelley (1974) suggest that collusion is less likely under product
differentiation, this is not confirmed by Dick (1996a). Collusion is
more likely in concentrated industries, according to Hay and Kelley
(1974) or Fraas and Greer (1977), but not according to Dick (1996a) or
Asch and Seneca (1976). And while Dick (1996a) finds a clear positive
link between capital intensity and the likelihood of collusion, Hay
and Kelley (1974) find no link between the ratio of fixed to total costs
and collusive conduct.17
A possible shortcoming of these studies is that the data either
originate from antitrust cases, and may therefore be subject to selec-
tion bias, or relate to export cartels, which cover only a very small
fraction of total economic activity. In contrast, the data on the British
cartels of the 1950s provide information which is both comprehen-
sive and, as argued above, relatively free from selection bias.

Data and Variables

I will use a simple but powerful way to model collusion in this


chapter, namely a dummy variable, COLL, which takes the value 1
for collusive industries and the value 0 for competitive industries.18
As pointed out above, an industry was classified as collusive in the

17. There is also an empirical literature on the related but different issue of cartel
duration (for instance, Jacquemin et al., 1981; Suslow, 1991; Dick, 1996b). Probably the
most robust result from this line of research is the negative effect of demand uncer-
tainty on cartel stability.
18. Going beyond this binary classification—for instance, by classifying industries with
respect to the ‘‘degree of collusion,’’—is not possible, given the information available.
The Competition Data 75

1950s if the products subject to significant restrictions accounted for


more than 50% of total industry sales revenue, and it was classified
as competitive if the products subject to significant or uncertain
restrictions accounted for less than 10% of industry sales revenue. All
other cases were excluded as ambiguous, leaving a sample of 156
industries.19
The concentration measure used here, C5, is the share of the five
largest producers in the total sales revenue of UK firms, excluding
firms with fewer than twenty-five employees. This is the measure
that is also used in later chapters. The data for 1958 (a year when the
restrictive agreements were still in place) were taken from summary
table 5 of the 1963 Census of Production, and correspond to Census
four-digit industries.20 This level of aggregation seems appropriate
for the analysis of collusive pricing.21 However, when competition
tends to operate at the regional rather than the national level, these
concentration ratios may be a poor measure of market structure. I
made an effort to identify such cases and exclude them from the
sample. Five industries were thus excluded, bringing the final sam-
ple size to 151.
For capital intensity or the level of sunk entry costs, I use two al-
ternative measures: the logarithm of the industry capital-labor ratio,
ln K=L, and the logarithm of the industry capital stock divided by the
number of plants in the industry, ln K=N. The former is a standard
capital intensity measure, while the latter is a measure of the abso-
lute capital requirements for setting up a plant of average scale. The
data on the value of the capital stock are estimates rather than pri-
mary data, and were taken from O’Mahony and Oulton (1990). See
chapter 4 for a detailed description of these data.
Although the capital stock data were at the three-digit level of ag-
gregation (which corresponds to broader industry definitions than

19. Since there were various possible reasons for classifying an industry as ambiguous,
it did not seem appropriate to regard these industries as intermediate with respect to
the state of competition and include them in the sample, except as a mere robustness
check. As discussed in Symeonidis (1999b), this has no effect on the results.
20. These are the only available concentration data for the 1950s that correspond to the
industry categories I will be using to analyze the effects of the 1956 legislation on market
structure in later chapters.
21. In some cases, a four-digit industry may comprise two or more submarkets which
are relatively independent on the demand side. I identified eight such cases in the
sample. Three were in the collusive group, and five in the competitive group, so it is
unlikely that any bias is caused from this factor.
76 Chapter 3

the four-digit level), it was often possible to derive reasonable ap-


proximations of capital stock at the four-digit level by using Census
data on the fraction of the three-digit industry investment accounted
for by each four-digit industry within any given three-digit industry.
In particular, I computed estimates of four-digit capital stock by
multiplying the 1958 three-digit capital stock by the ratio of four-
digit investment to three-digit investment, averaged over 1954 and
1958.
Data on investment, employment, and plant numbers were taken
from the industry reports of the Report on the Census of Production for
1958 (London: H.M.S.O., 1960–1963), and some adjustments were
made to ensure comparability in light of the fact that the O’Mahony
and Oulton figures are based on the 1968 Standard Industrial Clas-
sification, which is somewhat different from the 1958 S.I.C. Firms
employing fewer than twenty-five persons were not taken into
account.
Finally, I used advertising-sales ratios (ADS) and R&D-sales ratios
(RDS) to classify the industries in the sample. In particular, the
industries were split, on the one hand, into low-advertising and
advertising-intensive, using the 1% ADS as a cutoff point, and, on
the other, into low-R&D and R&D-intensive, using the 1% RDS as a
cutoff point. The variable AD1 was defined as a dummy variable
equal to 1 for industries with advertising-sales ratio higher than
1% and 0 otherwise. Similarly, the variable RD1 was defined as a
dummy equal to 1 for industries with R&D-sales ratio higher than
1% and 0 otherwise. The 1% cutoff point was chosen because it is
commonly used to classify industries according to advertising or
R&D intensity. In addition, it resulted in a not-too-unbalanced split,
whereas a higher cutoff point (2%, say) would leave only a small
number of industries in the advertising-intensive and R&D-intensive
categories. In any case, using the 2% cutoff point gave similar results.
Of the 151 industries in the sample, 111 had an ADS lower than 1%
and 119 had an RDS lower than 1%.
To construct these ratios for 1958, I combined data from various
sources, using an approach similar to that used in later chapters of
this book for constructing typical or average advertising-sales ratios
and R&D-sales ratios over the entire period 1954–1975. Details can
be found in chapter 4 and in Symeonidis (1999b).
There are several reasons why the advertising-sales ratios and the
R&D-sales ratios are used here merely to split the industries into
The Competition Data 77

groups rather than as continuous variables. First, data limitations


made it difficult to construct exact figures, especially for R&D inten-
sity and for advertising intensity in low-advertising industries. Sec-
ond, a similar approach is used in later chapters, where the analysis
is conducted separately for exogenous sunk cost, advertising-
intensive and R&D-intensive industries. And third, these binary
classifications reflect exogenous industry characteristics, such as
advertising effectiveness or technological opportunity, but are rela-
tively robust to other factors that may affect advertising or R&D in-
tensity. Since my primary purpose here is to study the links between
the incidence of collusion and some of its potential determinants, the
use of the dummy variables AD1 and RD1 is a way of minimizing
the problem of the potential endogeneity of advertising or R&D in-
tensity.22 Note that the fact that the results were similar when cutoff
points of 2% were used instead of 1% strengthens the case for treat-
ing AD1 and RD1 as exogenous, since it suggests that the results are
not driven by industries with intermediate advertising or R&D in-
tensity. They are, rather, driven by industries with either very low or
very high advertising/R&D intensity, and it is clear that very low or
very high advertising/R&D intensity is a largely exogenous feature of
industries.

Statistical Evidence

Descriptive statistics, separately for collusive and competitive in-


dustries, are given in table 3.3, and simple correlation coefficients
between COLL and each of the other five variables are given in table

22. The idea is that while the actual level of the advertising-sales ratio and the R&D-
sales ratio may depend on many variables, including the intensity of price competi-
tion, it is generally exogenous industry characteristics that will determine whether
these ratios are above or below 1% (or 2%). Thus, for an industry below the 1% cutoff
point, advertising/R&D is not an important strategic variable: in such an industry,
advertising is not very effective in raising consumers’ willingness to pay or there is
little scope for technological innovation from within the industry. On the other hand,
in an industry above the 1% cutoff point (and even more in an industry above the 2%
cutoff point), advertising/R&D ‘‘works.’’ Of course, whether such an industry has an
advertising-sales or R&D-sales ratio of 5% or 10% may be largely determined by
endogenous factors. However, my binary variables AD1 and RD1 are not very sensi-
tive to endogenous factors that affect advertising intensity and R&D intensity.
I should also point out that a comparison of advertising-sales ratios and R&D-sales
ratios across various years revealed very few instances where an industry had moved
from below 1% to above 1% or vice versa; and in most cases this was due to an exog-
enous institutional change, the introduction of TV advertising in the UK in the 1950s.
78 Chapter 3

Table 3.3
Descriptive statistics for collusive and noncollusive industries, 1958
Collusive Noncollusive
ðn ¼ 71Þ ðn ¼ 80Þ
C5 Mean 0.61 0.52
Standard deviation 0.20 0.28
Minimum 0.19 0.08
Maximum 0.98 0.99
ln K=L Mean 1.23 0.64
Standard deviation 0.73 0.86
Minimum 0.99 1.18
Maximum 3.10 2.38
ln K=N Mean 0.01 1.09
Standard deviation 1.15 1.33
Minimum 2.98 3.60
Maximum 2.17 3.42
No. of industries with AD1 ¼ 0 62 49
No. of industries with AD1 ¼ 1 9 31
No. of industries with RD1 ¼ 0 59 60
No. of industries with RD1 ¼ 1 12 20
Note: n indicates the number of industries.

3.4. There is a strong positive correlation between COLL and both


ln K=L and ln K=N, statistically significant at the 1% level, suggest-
ing that capital intensity and the level of setup costs are important
determinants of the likelihood of collusive pricing. There is also a
positive correlation between COLL and C5, which is significant at the
5% level but not at the 1% level. As can be seen from table 3.3, the
mean level of C5 is 0.61 for collusive industries and 0.52 for non-
collusive industries. Moreover, a test of independence of the respec-
tive means of C5 in the two groups rejects the null hypothesis of
independence at the 5% level although not at the 1% level.23 It
should be emphasized, however, that concentration is highly corre-
lated with both ln K=L and ln K=N in these data: the correlation co-
efficient between C5 and ln K=L is 0.53, and that between C5 and
ln K=N is 0.64. Thus it may be the case that the relatively higher level
of C5 in collusive industries simply reflects the higher level of capital

23. A Wilcoxon rank-sum test of the hypothesis that the distribution of C5 is the same
in the two groups also rejects the null at the 5% level but not at the 1% level.
The Competition Data 79

Table 3.4
Correlation coefficients between COLL and other variables, 1958
Correlation coefficient (significance level)
COLL and C5 0.19 (0.02)
COLL and ln K=L 0.34 (0.00)
COLL and ln K=N 0.40 (0.00)
COLL and AD1 0.29 (0.0002)
COLL and RD1 0.10 (0.23)

Table 3.5
Advertising, R&D, and collusive pricing, late 1950s
No. of No. of
collusive noncollusive
industries industries

Industries with ADS < 1% 62 49


Industries with 1% < ADS < 2% 6 14
Industries with 2% < ADS < 5% 3 6
Industries with ADS > 5% 0 11
Industries with RDS < 1% 59 60
Industries with 1% < RDS < 2% 5 11
Industries with RDS > 2% 7 9

intensity and setup costs in this group. I will return to this issue
below.
Note also the strong and highly significant negative correlation
between COLL and AD1, suggesting that collusion is less likely in
advertising-intensive industries than in industries with low adver-
tising intensity, as well as the negative relationship between COLL
and RD1, which is not, however, statistically significant at the 5%
or even the 10% level. A more detailed breakdown of industries ac-
cording to their competition regime, advertising intensity, and R&D
intensity is provided in table 3.5. Note that none of the industries in
the sample with ADS higher than 5% is collusive.
To check whether there is any (linear) association between con-
centration and collusion once we control for capital intensity, I have
also computed the partial correlation coefficient between COLL and
C5, keeping ln K=L constant. It is equal to 0.007, with a significance
level of 0.93. On the other hand, the partial correlation coefficient
between COLL and ln K=L, keeping C5 constant, is equal to 0.294,
80 Chapter 3

and is significant at the 1% level. Similar results are obtained when


using ln K=N instead of ln K=L. In other words, the association be-
tween collusion and concentration in these data virtually disappears
when one controls for capital intensity or setup cost, but the associ-
ation between collusion and capital intensity or setup cost remains
strong even when one controls for concentration.
These results are confirmed in Symeonidis (1999b), where the anal-
ysis presented here is extended in several ways. In particular, I
have run regressions of COLL on C5, AD1, RD1, a measure of de-
mand growth, and a measure of capital intensity or setup cost
(ln K=L or ln K=N), and I have also allowed for a possible nonlinear
effect of C5 and the growth variable on COLL. In some specifications
I have included among the regressors a proxy for the degree of for-
eign competition, based on effective rates of protection across indus-
tries. I have also included among the regressors a set of sector
dummies as a partial check for misspecification due to omitted vari-
ables or the presence of industry effects. I have estimated the model
using both a standard one-step probit specification and a two-step
procedure meant to control for the endogeneity of concentration. I
have included a producer-good/consumer-good dummy among the
regressors or alternatively I have run regressions separately for pro-
ducer-good and consumer-good industries. I have also estimated the
model using a slightly smaller sample, excluding industries with a
high value of imports relative to sales by domestic firms, to confirm
that no measurement error is caused by the fact that the concentra-
tion ratios are not adjusted for imports.
Finally, and perhaps most important, I have experimented with
several different definitions of the variable COLL, using modified
criteria to classify the industries as collusive or noncollusive, in an
attempt to check for measurement error or selection bias in the
classification of industries. For instance, in one of these experiments
I have treated all ambiguous industries as intermediate, and in an-
other I have classified as collusive those industries which did not
register any agreements but are reported as having been collusive or
potentially collusive in the Political and Economic Planning survey
or the Board of Trade annual reports.
The results were robust to all these extensions and can be briefly
summarized as follows. The strong positive association between
capital intensity or the level of setup costs and collusive pricing is
confirmed throughout. On the other hand, no clear link between
The Competition Data 81

concentration and the likelihood of collusion can be found. In par-


ticular, there is no evidence of any linear association, although a non-
linear relationship is present in some regressions, with both very
low and very high concentration hindering collusion. There is strong
evidence that collusive pricing is less likely in advertising-intensive
industries than in low-advertising industries, and weak evidence
that it is less likely in R&D-intensive industries than in low-R&D
industries. Some of the results indicate that collusion is more likely
under conditions of moderate market growth than in a market with
declining or stagnant demand, and less likely under fast growth than
under moderate growth. Finally, my proxy for the extent of foreign
competition is negatively associated with collusion in many regres-
sions, but the effect is relatively weak and not statistically significant
once sector dummies are included among the regressors.24

Interpreting the Evidence

A possible reason for the absence of any clear link between concen-
tration and the incidence of collusion in the present context is that
the coordination and monitoring of collusion must have been greatly
facilitated by the fact that the agreements were not illegal and were
often operated by members of trade associations. In fact, the admin-
istration of collusive agreements was one of the primary functions of
British industrial trade associations in the 1950s, and often their main
function (see Political and Economic Planning 1957). On the other
hand, a nonlinear relationship between concentration and collusion
could result from the fact that high levels of concentration are often
associated with significant firm asymmetries or the presence of dom-
inant firms, and these factors may hinder collusion.25

24. Thus, although there is no reason to doubt that the growth of tariffs in the 1920s
and the 1930s had greatly facilitated the operation of price agreements, as argued by
Swann et al. (1974, p. 197), it is not clear whether tariff protection was an important
determinant of the incidence of collusion across industries in the 1950s.
25. One argument which was used by the authors of the 1946 Board of Trade report on
British cartels to explain the absence of restrictive agreements in several industries
where a single firm controlled a very large part of the market is that in these industries
there may be little scope for collusion, since prices are effectively decided by the
dominant firm. However, such cases are not very numerous at the four-digit industry
level—although they may be more numerous at a lower level of aggregation. More-
over, some industries with dominant firms are not included in the present data set,
because their concentration ratio is not reported in the official statistics to avoid dis-
closure of information about specific firms.
82 Chapter 3

Some remarks on the observed weak negative link between R&D


intensity and collusion are also in order, especially when this is com-
pared against the much stronger negative link between advertising
intensity and collusion. There are at least two factors that may help
explain why the negative link is not stronger for R&D intensity as
well. First, a potentially important institutional factor is that public
procurement procedures in the UK in the 1950s did little to deter
collusion in some R&D-intensive industries, perhaps because their
main objective was not to promote competition but to maintain a
stable supply by means of a smooth sharing of orders among estab-
lished firms in these industries.
A second explanation derives from models of cartel stability which
predict that quality differences between firms in vertically differen-
tiated markets hinder collusion. Large differences in product quality
between firms in R&D-intensive industries are probably not as com-
mon as large differences in brand image between firms in advertising-
intensive industries. This is because in an R&D-intensive industry, a
low-quality firm may find it difficult to compete with a high-quality
rival, while in an advertising-intensive industry, heavy advertisers
coexist with firms that do not advertise but employ other market-
ing strategies.26 Thus collusive pricing should be easier to sustain in
R&D-intensive industries than in advertising-intensive industries.
There is in fact some case-study evidence on the British cartels
which seems to be consistent with the above interpretation. In some
R&D-intensive industries, such as heavy electrical machinery or tele-
communications equipment, the collusive agreements were mostly
operated by a small number of R&D-intensive firms, which were the
only UK producers of the products in question. Hence price collusion
was presumably not hindered by large quality differences. More-
over, in several other industries the agreements involved ‘‘patent
pooling’’ and exchange of technical information (but not coopera-
tion in R&D). These schemes may have been used to limit quality
differences and ensure the success of the pricing arrangements.
Finally, it must be acknowledged that the use of a cutoff point such
as an RDS of 1% or 2% to classify industries may not be sufficient for
capturing the links between collusion and technological opportunity

26. Differences in product quality between firms will also be small in industries with
process rather product R&D. This may also weaken any link between R&D intensity
and collusion, although it needs to be said that several industries where process R&D
is important, such as basic chemicals, are not included in the sample used here.
The Competition Data 83

across industries. As will be discussed in detail in chapter 6, there is


indeed evidence of a strong negative relationship between collusion
and the number of innovations produced across UK manufacturing
industries within the RDS > 1% class during the 1950s.
The two most important factors that facilitated explicit collusive
pricing agreements in UK manufacturing during the 1950s are low
advertising effectiveness and high capital intensity or setup costs.
The observed negative link between advertising and the likelihood
of collusion is consistent with models of cartel stability that em-
phasize vertical product differentiation, and also with key insights
from the literature on coordination, as opposed to enforcement, of
collusive agreements. Note that this negative link cannot be due to
a negative effect of collusion on advertising. In chapter 5 I will show
that the intensification of price competition following the abolition
of cartels in the UK caused advertising intensity to fall. Thus the
negative association between advertising and collusion in the 1950s
that I find in the present chapter can only mean that collusion is
less likely to occur if advertising intensity (and hence advertising
effectiveness) is high.
The positive relationship between capital intensity or setup costs
and the incidence of collusion is of particular interest because the
role of these variables has not been much emphasized in the theo-
retical literature on collusion. A possible objection to this finding
is that capital intensity and the level of sunk costs, both of which
are inevitably measured here with some error, may in fact be picking
up the effect of concentration in the present sample. To assess the
validity of this objection, I have tried using three-digit (unadjusted)
ln K=L and ln K=N. The statistical results were not significantly af-
fected, which implies that they are generally robust to more or less
measurement error in these variables. These results suggest that the
ease of entry into industries is also a key factor—together with in-
dustry characteristics that affect coordination and the enforcement of
collusive agreements—in the formation and stability of cartels.

3.5 Concluding Remarks

This chapter has prepared the way for the analysis of the effects
of the 1956 legislation in the rest of the book. In the first part, I have
described the available data on cartels and competition in British in-
dustry from the 1950s to the mid-1970s. I have argued, on the basis of
84 Chapter 3

extensive indirect evidence, that there should be no significant mea-


surement error or selection bias in the competition data, and I have
also emphasized that any such data problems would not create seri-
ous difficulties for the analysis of the effects of the 1956 Act, since the
key distinction that needs to be made when carrying out this analy-
sis is between industries with a change of competition regime and
industries without such a change. Finally, I have described in detail
my methodology for constructing these two groups of industries.
I have then discussed, in the second part, some statistical evi-
dence on the factors facilitating or hindering explicit collusive pricing
across British manufacturing industries during the 1950s. The evi-
dence suggests that the most important factors that facilitated collu-
sion were low advertising effectiveness and high capital intensity or
setup costs. Let me now conclude this chapter by pointing out some
of the implications of this statistical evidence for the rest of the book.
A key implication is that the group of industries with a change in
competition regime after 1958 and the group without such a change
did not differ significantly with respect to their market structure
prior to the introduction of the legislation. This lends some support
to a central assumption of this book, namely, that any difference in
the evolution of market structure between the two groups after 1958
can be attributed to the effect of the 1956 legislation rather than to
any unobserved characteristics that may be correlated with both con-
centration and collusion.
This argument should not be overemphasized, however. The rea-
son is that the similarity of initial conditions is neither a necessary
nor a sufficient condition for establishing the validity of the differ-
ence-in-differences methodology in the present context. This is be-
cause what matters most is trends, not levels; in other words, the
methodology is valid as long as there are no unobserved character-
istics that differ between the two groups of industries and cause
them to evolve in different ways during the period under study. The
fact that the two groups did not differ significantly with respect to
their initial market structure conditions is only an encouraging indi-
cation in that respect. At the very least, it may help dispel a potential
concern with the present research design, namely, the possibility that
some of the difference that we observe between the two groups after
1958 is due to purely statistical factors (i.e., the simple fact that the
scope for an increase in concentration is larger, the lower the initial
level of concentration in an industry). Further support for the validity
The Competition Data 85

of the difference-in-differences methodology will be provided in later


chapters by an analysis of the evolution of market structure in the two
groups of industries before the effective implementation of the 1956
legislation.
On the other hand, the negative association between advertising
intensity, and possibly also R&D intensity, and collusion may seem,
at first sight, to complicate the interpretation of some of the results of
the next four chapters. However, this is not the case. First, most of
the analysis of the effects of the 1956 Act will be performed separately
for exogenous sunk cost, advertising-intensive, and R&D-intensive
industries. Second, it will be confirmed that these associations are due
to more fundamental correlations between the propensity to collude
and exogenous industry characteristics, such as advertising effective-
ness and technological opportunity, which can be largely controlled
for in regression analysis using panel data. And third, further support
for the key maintained assumptions of the present study will be
derived in later chapters from an analysis of the evolution of adver-
tising intensity and innovative output in the two groups of industries
before the effective implementation of the 1956 legislation.
This page intentionally left blank
4 Price Competition and the
Evolution of Concentration
in Exogenous Sunk Cost
Industries

4.1 Introduction

In this chapter I begin the analysis of the effects of the intensification


of price competition in British industry following the 1956 cartel
legislation. This chapter focuses on the effect of competition on mar-
ket structure in exogenous sunk cost industries. In a seminal paper,
Selten (1984) predicted that a switch from collusive to noncollusive
behavior caused by a toughening of competition policy would cause
a decrease in the number of firms in a homogeneous-good industry.
The reason for this result is that the number of firms is determined
by a free-entry condition which requires that net profit in long-run
equilibrium be driven to (almost) zero by entry irrespective of firm
conduct. A shift in the competition regime that causes the gross profit
of each firm to fall, given the initial industry structure, triggers a
process of exit and merger through which gross profit rises again to a
level that covers sunk costs. In other words, the zero-profit condition
is restored by means of a restructuring of the industry. This frame-
work has been enriched in Sutton’s (1991, 1997a, 1998) comprehensive
theory of market structure, which distinguishes between exogenous
sunk cost, advertising-intensive, and R&D-intensive industries, and
examines the links between market size, sunk costs, and concentra-
tion in the three types of industries.
This chapter sets out to test the Selten-Sutton prediction of a posi-
tive effect of price competition on concentration in exogenous sunk
cost industries. I begin by presenting a simple theoretical framework
for analyzing competition in exogenous sunk cost industries based
on Selten (1984) and Sutton (1991), and I also illustrate the results
using a specific theoretical model. This is followed by a detailed dis-
cussion of the data set used for the econometric analysis. Since many
of the variables used here will also be used in later chapters, this
88 Chapter 4

discussion serves as a reference for later chapters as well. Finally,


I present the econometric results. They suggest that firm conduct,
which is partly determined by exogenous institutional factors, is an
important determinant of market structure.
While the idea that more intense price competition might lead to
higher concentration is certainly not new, the empirical evidence on
this issue has so far been inconclusive. Early attempts to capture this
effect by introducing lagged profitability as a regressor in empirical
models of the determinants of concentration change have produced
mixed results (see Martin 1979a, 1979b). More recently, the literature
on economic integration has found some evidence of a positive effect
of a reduction of trade barriers on concentration (for example, Caves
1988, Sleuwaegen and Yamawaki 1988). However, since economic
integration involves more than just an intensification of price com-
petition, it is not easy to draw clear implications for the competition-
concentration relationship from these results.1 A similar difficulty
arises with respect to studies of the effects of deregulation and liber-
alization on industrial structure and performance: these policy inter-
ventions typically involve a reduction or removal of entry barriers as
well as an intensification of competition, and it is not easy to separate
the two effects.
Studies of the impact of cartel policy on the structure of industry
should be more informative in this respect. In Britain, the argument
that the breaking up of cartels might lead to higher concentration
was used in some Restrictive Practices Court cases by associations
defending their agreements and became a topic of debate during the
early discussions of the effects of the 1956 Act. However, little evi-
dence was available at the time either for or against this view (Wilson
1962). In later years, the hypothesis of a causal link between the in-
troduction of cartel laws and a rise in concentration was also
advanced by economic historians who claimed that differences in the
evolution of market structure between the US and the UK could be
partly attributed to differences in the competition policy regimes in
the two countries (Hannah 1979, Freyer 1992).
In particular, Hannah and Freyer argued that the prohibition of
cartels in the US in the 1890s was a key factor behind the great merger

1. It is equally difficult to interpret the evidence of a positive link between price com-
petition and concentration in US exogenous sunk cost industries reported by Robinson
and Chiang (1996), since their measure of competition was based partly on demand
characteristics, such as the degree of product homogeneity, and partly on very specific
factors that were assumed to affect firms’ conduct, such as the infrequency of orders.
Price Competition and Concentration 89

wave of the years 1898–1902. This is a view shared by a number of


American historians but rejected by others, for it has proved difficult,
in detailed empirical studies, to isolate the effect of competition policy
from other influences on mergers, such as the rise of large-scale pro-
duction, changes in corporation law, or the growth of the stock market
(see Nelson 1959, Bittlingmayer 1985). Hannah and Freyer contrasted
the US experience with that of the UK, where cartel policy was per-
missive and market structure remained relatively fragmented until
the 1960s. After that date the reversal of cartel policy was associated
with the great UK merger wave of the 1960s that closed the gap in
industrial organization between the two countries. However, Hannah
has also warned of the potential dangers of such broad comparisons
by highlighting the case of Germany, where, despite the significant
increase in the degree of cartelization in the early twentieth cen-
tury, the predominance of large firms was more noticeable than in
Britain.
More systematic evidence on the impact of the 1956 Restrictive
Trade Practices Act on the evolution of market structure in the UK
was examined in two studies undertaken in the 1970s. Like the pres-
ent book, these studies were based on a comparison of industries
affected by the 1956 legislation and industries not affected (without
distinguishing between exogenous sunk cost, advertising-intensive,
and R&D-intensive industries). However, they were both subject to
methodological problems and produced conflicting results.
The study by Elliott and Gribbin (1977) is the one closer to the
present one, in terms of both the research design and the empirical
results. These authors examined restrictive agreements of nation-
wide application between manufacturers registered before 1960. They
found that the rise in the five-firm sales concentration ratio between
1958 and 1968 was, on average, significantly larger in industries with
terminated agreements than in industries without agreements, and
that most of the differential effect occurred between 1963 and 1968.
They also noted, however, that market size was growing faster for
industries in the latter group throughout the period, so the cause of
the observed difference in the evolution of concentration between the
two groups is not clear. Moreover, the criteria they used to classify
industries to one or the other group were not made explicit, and no
attempt was made to use additional data sources in order to identify
unregistered agreements or cases of detected collusion after 1960.
On the other hand, O’Brien et al. (1979) examined merger activity
over the period 1959–1972 for a sample of about thirty industries
90 Chapter 4

split into four groups: a group with agreements upheld by the Court,
another with agreements struck down by the Court, a third group
comprising industries which voluntarily abandoned their agreements,
and a control group of industries which, according to the authors,
had not been subject to collusion prior to the introduction of the 1956
Act. Three different measures of merger activity were used: the ratio
of expenditure on acquisitions to total assets, the proportion of firms
in an industry taking over other firms, and the proportion of firms in
an industry being acquired. The authors found no statistically signif-
icant difference in the ratio of expenditure on acquisitions to total
assets among the four groups, and they also found that the control
group had the highest proportion of both acquirers and acquired
firms. They concluded that there was no evidence of any effect of the
1956 Act on merger activity during the period under study. How-
ever, their sample may have been too small to be representative.
Also, the criteria used to classify industries to the four groups were
not always clear.2
In summary, the empirical evidence from previous studies of the
effect of competition on concentration in the UK, and also in the US,
has been inconclusive or mixed. This chapter undertakes a systematic
and extensive reexamination of the UK evidence, focusing on exoge-
nous sunk cost industries. Later chapters in the book will examine
the evidence for advertising-intensive and R&D-intensive industries.
Overall, I extend the previous studies of the effects of the 1956 Act on
market structure in several ways. First, I relate my empirical analysis
to a theoretical framework, two key elements of which are the dis-
tinction between classes of industries and the focus on jointly deter-
mined endogenous variables. Second, I look at the impact of the Act
over a longer time period, from the mid-1950s to the mid-1970s, and
for the whole of manufacturing industry. Third, I use more compre-
hensive data on competition; in particular, I use information from
several sources to identify unregistered agreements and cases of col-
lusion in the 1960s. Fourth, I use explicit criteria to classify industries
according to their competitive status. Fifth, I pay much attention to
the issue of potential selection or endogeneity bias when interpreting
the evidence. And, finally, I perform regression analysis to control

2. For instance, some of the industries included in the control group had in fact been
subject to restrictive agreements in the 1950s, and one of the industries included in the
group whose agreements were upheld by the Court dropped the agreement shortly
after the Court’s decision.
Price Competition and Concentration 91

for other influences on the evolution of market structure, firm con-


duct, and performance in the UK between the 1950s and the 1970s.

4.2 Theoretical Framework

Consider first an exogenous sunk cost industry producing a homo-


geneous product. A two-stage game can be used to model competi-
tion in such an industry. There are N0 potential entrants. At stage 1
these firms decide whether or not to enter the industry, and those
that do enter, pay an exogenously given sunk cost of entry f . We
assume that N0 is sufficiently large so that at any equilibrium of the
game there is at least one nonentering firm. At stage 2 those firms
that have entered simultaneously set prices or quantities.
The notion of an exogenous sunk cost industry should be seen as a
useful approximation. It attempts to capture two ideas. The first is
the idea that firms must operate plants at a level close to minimum
efficient scale to be able to compete effectively in the long run. It is in
this sense that the setup cost f , defined as the cost of acquiring a
plant of minimum efficient scale, net of resale value, is assumed to be
exogenously determined by technology. This is, of course, a simpli-
fication, for it is well known that plants of suboptimal size can often
coexist in an industry along with more efficient plants (see Scherer
et al. 1975). Still, the average or typical plant size or capacity in an
industry will be largely determined by technology. Hence the cost
of successful entry will largely depend on exogenously determined
characteristics of the production process.
The second thing that the notion of an exogenous sunk cost in-
dustry attempts to capture is the idea that in many industries there is
limited scope, at any given time, for investment to decrease cost or to
enhance quality or brand image. For this reason, the setup cost f is
assumed to be the only sunk cost in the two-stage game above. Again
this is an approximation, since endogenous sunk costs are unlikely to
be zero in any industry.3 What the structure of the two-stage game
tries to capture is that these endogenous sunk costs are very low in

3. Advertising and R&D are probably not the only endogenous sunk costs incurred by
firms. For instance, the cost of developing a network of contacts or relationships may
also be thought of as an endogenous sunk cost. The reason for ignoring it in the pres-
ent book is twofold. First, it is probably less important (or provides less scope for
endogenous investments by firms) than advertising or R&D, at least in the manu-
facturing sector. Second, it is difficult, if not impossible, to measure in a systematic
way across industries.
92 Chapter 4

industries where advertising or R&D is not an important strategic


variable, and that this is something largely determined exogenously:
advertising effectiveness and technological opportunity are simply
very low in certain industries.
In other words, while the general model for analyzing the links
between price competition, firm strategy, and market structure must
necessarily allow for endogenous sunk costs, there is a special limit-
ing case within that general model which arises when advertising
effectiveness and technological opportunity are both very low. The
equilibria of this limiting-case model coincide with the equilibria of
an elementary model in which there is no advertising or R&D. That
is the model of the present chapter.

The Second-Stage Subgame

The equilibrium outcome of the second-stage subgame can be repre-


sented by a vector of (gross) profits pi ðS; N; t; c1 ; . . . ; ci ; . . . ; cN Þ, where
S is market size, an exogenous demand-shift parameter; N is the
number of firms that have entered at stage 1; ci is a vector of param-
eters specific to firm i (such as marginal cost, capacity, the number of
plants operated, and so on); and t is a measure of the intensity of
price competition. In particular, t captures the idea that for any given
number of firms N, pi will depend on the firms’ pricing behavior,
which in turn will depend partly on exogenous institutional factors,
such as the climate of competition policy or the degree of economic
integration.
The above interpretation of t implies that this variable can be
thought of as an inverse measure of the ‘‘degree of collusion.’’ A
well-known result in oligopoly theory states that under certain con-
ditions, any individually rational and feasible payoff vector can be
sustained as an equilibrium of an infinitely repeated game if the
players are sufficiently patient (Fudenberg and Tirole 1991). It seems
natural, then, to assume that the climate of competition policy or other
institutional factors will considerably affect the probability of any
particular outcome being realized.4 For instance, making collusion
illegal will make coordination between firms more difficult and will

4. Alternatively, one could assume that firms always achieve the highest level of col-
lusion that is sustainable, given a number of parameters taken as exogenous at the
‘‘short-run’’ competition stage. Under this interpretation, a rise in t might correspond
to a lower critical discount factor in an infinitely repeated game.
Price Competition and Concentration 93

also increase the expected costs of collusion. It should be stressed,


however, that, since t refers to the type of pricing conduct, it is not
equivalent to the price-cost margin, which is a measure of perfor-
mance and can only be treated as endogenous.
Consider now the benchmark case of symmetric single-plant firms.
The equilibrium payoff of firm i can be written as pi ðS; N; tÞ. Alter-
natively, one can define a concentration measure C, whose value
increases in 1=N, such as the concentration ratio, and write gross
profit as pi ðS; C; tÞ. We assume that qpi =qS > 0 and qpi =qt < 0, that is,
the equilibrium gross profit of firm i increases with market size and
decreases when price competition intensifies (t rises). We also assume
that pi is decreasing in the number of firms N and increasing in the
concentration measure C, which is in fact the standard prediction of
the traditional structure-conduct-performance approach.5
These assumptions are quite plausible, especially in a situation
where the total sales revenue in the industry depends mainly on
market size S rather than N or t, and the marginal cost curve is rela-
tively flat and not very much affected by S, N, or t.6 Note also that
the key assumption of a negative effect of N on profit per firm is

5. According to this approach, concentration increases profitability because it facili-


tates the exercise of market power by firms; moreover, profits are sustained through
‘‘barriers to entry’’ such as economies of scale, product differentiation through adver-
tising or R&D, and so on. This view has been subject to two main criticisms. One crit-
icism, due to Demsetz (1973), is that high profitability and high concentration in an
industry may both be caused by large efficiency differences between firms: since effi-
cient firms have large market shares and make high profits, concentration and industry-
level profitability will be correlated even though there will be no causal link between
the two. Another criticism is that concentration is itself affected by firm conduct, for
instance, by the intensity of competition and by firms’ strategies regarding advertising
and R&D (which therefore should not be regarded—as is typically the case within the
structure-conduct-performance approach—as exogenous barriers to entry).
The view taken in this book is that the structure-conduct-performance approach,
although not fundamentally wrong, is incomplete, since it neglects or downplays the
endogeneity of market structure and the existence of feedback effects from conduct
and performance to structure. The present theory embeds the key result of the structure-
conduct-performance approach, namely, the positive effect of concentration on profit
margins, in a broader framework that explicitly treats market structure as endogenous.
Moreover, the present theory is not inconsistent with the efficiency differences ap-
proach, as will be explained below.
6. For instance, in the case of constant marginal cost and unit-elastic demand, these
assumptions can be reformulated in terms of some very plausible hypotheses involv-
ing price p. In particular, a sufficient condition for qpi =qS > 0 is qp=qS b 0; a sufficient
condition for qpi =qt < 0 is qp=qt < 0; and a sufficient condition for a negative relation-
ship between pi and N is that price is nonincreasing in N. Note that this latter condition
has received considerable theoretical and empirical support—although the effect of
94 Chapter 4

satisfied under fairly general demand and cost structures in many


oligopoly models, including the Cournot model (see Seade 1980) and
simple models of repeated games (in the latter case the result is for
the maximum sustainable collusive profit).

The Symmetric Equilibrium and Comparative Statics

At stage 1 of the game, firms enter as long as the profit they expect to
make at stage 2 covers the cost of entry. Hence the equilibrium level
of N is given, in the symmetric case, by the largest integer satisfying

pi ðS; N; tÞ b f ; Ei: ð4:1Þ


Note that the free-entry condition (4.1) is assumed to hold, whatever
the intensity of price competition in the final stage of the game. This
is a key feature of the model, and it is consistent with the empirical
evidence on the British cartels described in chapter 2. We further as-
sume that the equilibrium level of N, as defined by (4.1), is unique;
this will be the case if the average cost curve is either U-shaped or
everywhere declining.
It is then easy to see that an increase in the intensity of competition
t will generally cause a fall in the equilibrium number of firms N 
and a rise in equilibrium concentration C  : gross profit, which has
fallen to a level lower than f following the increase in t, can be restored
to a level at least as high as f only through a fall in N by way of exit
or mergers. Note that the free-entry condition (4.1) also implies that
an increase in market size S or a fall in the setup cost f will cause a
rise in N  and a fall in C  .

Extensions

Next, consider an exogenous sunk cost industry producing a hori-


zontally differentiated product. Note that this includes the case of

entry on price may often be modest (see Bresnahan and Reiss 1991, Geroski 1991,
Sutton 1997a).
To see this, define S ¼ Npq, where q is output per firm, and notice that firm profit
can be written as p ¼ ð p  cÞq ¼ S=N  cq, where c is marginal cost. Then suppose that
N rises, which implies that pq falls, given S. Now if p is nonincreasing in N, then p
must fall when N rises, whatever happens to q: if q falls, p ¼ ð p  cÞq also falls, while
if q rises, p ¼ S=N  cq again falls. A similar argument can be made with respect to
changes in S. As for t, notice that a rise in t (i.e., a movement away from joint profit
maximization and toward the one-shot Nash equilibrium) causes p to fall and q to rise
so that pq remains constant, given S and N; hence p ¼ S=N  cq falls.
Price Competition and Concentration 95

spatial differentiation. The structure of the game is the same as in


the homogeneous-good industry case. In the benchmark case of
symmetric single-plant, single-product firms, the equilibrium gross
profit of firm i can be written as pi ðS; N; h; tÞ, where h is a measure of
the (exogenous) degree of horizontal product differentiation, with
qpi =qh > 0.7 The equilibrium number of firms N  is again the largest
integer satisfying the free-entry condition
pi ðS; N; h; tÞ b f ; Ei: ð4:2Þ

Assuming, as in the homogeneous-good industry case, that pi is


decreasing in N, qpi =qS > 0, and qpi =qt < 0, it is easy to check that N 
is decreasing in t and f and increasing in S and h.8
The above framework can be easily extended to exogenous sunk
cost industries consisting of multiplant firms, or where the average
cost curve becomes horizontal after a certain production level, or
where some or all firms produce more than one variety of the (dif-
ferentiated) product. The only difference is that there now exists a
multiplicity of equilibria. The most fragmented equilibrium is the
one where each firm operates only one plant, or produces at mini-
mum efficient scale, or produces one variety of the product, whereas
more concentrated equilibria, symmetric or asymmetric, can occur
if some or all firms operate more than one plant, or produce at a
level higher than minimum efficient scale, or produce more than one
variety of the product.
Whether the structure of an industry is more fragmented or more
concentrated in the presence of multiple equilibria will depend on

7. Note that h has a direct equivalent in standard models of horizontal product dif-
ferentiation: it would be the inverse of the degree of product substitutability in a non-
address model of product differentiation, such as Dixit and Stiglitz (1977), or the
consumers’ unit transport cost (or intensity of preferences) in an address model of
product differentiation without choice of location, such as Salop (1979). Both these
models can be seen as special cases of the general framework presented here, since h
is taken in these models as exogenous and pi is increasing in h. On the other hand,
there is a sharp distinction between horizontal product differentiation and advertising
in the present theory—unlike much of the literature following the structure-conduct-
performance approach, in which advertising was very often used as a proxy for the
degree of product differentiation. Advertising is modeled in chapter 5 as a source of
vertical (as opposed to horizontal) product differentiation.
8. See Norman and Thisse (1996) and d’Aspermont and Motta (2000) for an analysis of
the positive effect of price competition on concentration using address models of hor-
izontal product differentiation. Section 4.3 below illustrates the positive competition-
concentration link in the context of a nonaddress model of product differentiation.
96 Chapter 4

various factors that are difficult to observe or measure across indus-


tries. For instance, the presence of economies of scope in setup costs
or of first-mover advantages will favor more concentrated equilibria.
Note that more concentrated equilibria can also occur with single-
plant, single-product symmetric firms with declining average cost
curves when part of the fixed cost is incurred at the price competi-
tion stage—that is, only if the firm produces positive output (Vickers
1989). Finally, more concentrated equilibria can also occur when the
zero-profit condition is violated because of the existence of institu-
tional or other barriers to entry or the use of entry-deterring strat-
egies by incumbent firms (Lyons and Matraves 1996). In all these
cases, (4.1) and (4.2) define the minimum equilibrium level of con-
centration as a function of t and the other exogenous variables.
In conclusion, the present theoretical framework provides clear
predictions for exogenous sunk cost industries. The next section illus-
trates these results using a specific oligopoly model. Before analyz-
ing this example, however, several remarks on the general model
are in order.

Remarks and Clarifications

To begin with, the use of a two-stage game to model what is essen-


tially a dynamic process of repeated interaction may be seen as a lim-
itation of the present theory. However, this formulation captures the
key distinction between short-run and long-run decisions and the
notion that long-run decisions constitute a commitment at the time
when short-run decisions are being taken. Hence, at any point in
time some costs can be thought of as sunk, even though in practice
firms incur a continuous stream of expenditures.
It should also be stressed that the capital stock (or any other
firm-specific asset created through sunk expenditures) depreciates
and has to be renewed. This implies that the mechanism driving the
results of the present model—namely, the requirement that gross
profit be sufficiently high to cover sunk costs—is entirely appro-
priate for analyzing the effects of the 1956 Restrictive Trade Prac-
tices Act. In other words, the fact that the capital stock was already
in place when the exogenous change in the institutional framework
(and thus the increase in the intensity of price competition t) oc-
curred in the UK is irrelevant for the long run, because the capital
Price Competition and Concentration 97

stock has to be replaced periodically. It is, of course, relevant for the


short run, for it might imply a relatively weak short-run competition
effect on concentration and a relatively slow adjustment to the new
long-run equilibrium.9
Second, it may be asked why collusion should occur if profits are
to be driven to zero by free entry. This may be easier to justify in a
context with asymmetric firms, since in that case it is only the mar-
ginal firm that makes zero profit. Hence the low-cost firms may gain
from collusion in the long run, even though the high-cost firms’
profits will be driven to zero. In fact, the evidence from the UK car-
tels that I have discussed in chapter 2 suggests that price determi-
nation was often a bargain between the more efficient and the less
efficient cartel members, with prices set at a level that allowed the
high-cost firms to break even.
An alternative justification, which is also valid for the case of sym-
metric firms, is that there are short-run gains from price collusion
between a given number of firms, even though these gains are
eventually eliminated by changes in market structure. Even when
collusion is illegal, these gains may well compensate for any poten-
tial losses in case firms are prosecuted for violating the cartel laws,
especially if collusion is tacit and therefore difficult to detect and not
necessarily, on its own, in violation of the law. In the present context,
of course, collusion was not a violation of the law because cartels
were not illegal in the UK in the 1950s. Moreover, once price collu-
sion is established, it may persist, even though profits will eventually
be driven to zero by entry, because its breakdown will result in
short-run losses. The key point in these arguments is that entry and
exit do not occur as fast as price or output changes, a feature cap-
tured by the model presented here.
Third, the theoretical predictions obtained in this section relate to
the minimum rather than the actual level of concentration. Sutton

9. The adjustment need not be slow if the cost of the capital investment is spread
across many years through interest payments or depreciation allowances. Mergers
intended to restore the market power of firms following a rise in the intensity of com-
petition could also speed up the adjustment of concentration to its new long-run equi-
librium value. Another factor that could speed up the adjustment is the fact that more
efficient firms may acquire less efficient ones soon after competition intensifies to avoid
pronounced price-cutting behavior. In turn, less efficient firms, which feel vulnerable
once competition intensifies, may prefer to be acquired immediately (i.e., before com-
petition seriously affects their balance sheet position).
98 Chapter 4

(1991, 1998) has proposed a ‘‘bounds’’ approach for testing this class
of predictions.10 However, it is difficult or inappropriate to apply a
bounds approach in the present empirical context, for reasons that
will be discussed later in this chapter. The alternative is to carry out a
regression analysis of the competition effect on market structure, on
the implicit assumption that the theoretical results obtained for
minimum concentration apply to actual concentration as well. In
other words, it will be assumed that the mechanism identified
for the benchmark case of symmetric single-plant, single-product
firms dominates any other effects that may arise in a more complex
setting.11
Clearly, the validity of such a claim must be tested against the em-
pirical evidence. However, one remark may be made here. The key
issue is whether the symmetry assumption is appropriate in light
of the existence of considerable asymmetries between firms in any
given industry. These asymmetries often reflect efficiency differ-
ences. Now one might expect that once price competition intensifies,
high-cost firms (which were previously making zero or almost zero
net profit) will be more easily driven out or taken over, while low-
cost firms (which were previously making positive net profit) will be
more likely to stay in the industry and expand through growth or
mergers. At the new long-run equilibrium, the marginal firm will
again be making zero or almost zero net profit. Hence, in the ab-
sence of any initial negative correlation between firm size and
efficiency—which seems quite implausible for the typical industry
—concentration should rise. In particular, the expansion of low-cost
firms will only reinforce the effect of falling firm numbers on con-
centration. This story is perfectly consistent with the basic intuition
from the symmetric case. In fact, the model presented in this section
captures the basic mechanism driving the evolution of market struc-

10. The bounds approach, as described in Sutton’s work, has mostly focused on ana-
lyzing the properties of lower bounds to concentration. Recent theoretical work by
Nocke (2000) has extended this approach to the study of upper bounds to concentration
in different classes of industries.
11. A more complex theoretical framework for the analysis of market structure would
combine deterministic factors deriving from game-theoretic models of entry (such as
the one described in this chapter) with stochastic patterns of firm growth that may ac-
count for much of the asymmetry observed in real markets (such as those analysed in
Hart and Prais 1956, Simon and Bonini 1958). Sutton (1998) presents an integrated
theory of the determinants of market structure that combines the two approaches.
Price Competition and Concentration 99

ture in exogenous sunk cost industries, while abstracting from com-


plications that arise from firm heterogeneity.12
Finally, the assumption of the exogeneity of the intensity of price
competition t is clearly a simplification. However, it is not unreal-
istic, even on theoretical grounds alone, given the well-known mul-
tiplicity of equilibria in models of infinitely repeated games. More
important, this assumption is justifiable in the present empirical
context for two reasons. First, the key determinant of changes in t
during the period under study was an exogenous change in the
institutional framework, namely, the introduction of cartel law. In
particular, it was shown in chapter 2 that the intensity of price
competition increased, following the abolition of price-fixing, in
the large majority of the previously cartelized industries. Of course,
whether t would change at all depended also on its initial value. The
second point therefore is that, as shown in chapter 3, cartelization in
the 1950s was largely a function of exogenous industry character-
istics rather than of endogenous variables such as concentration. In
other words, although in principle the intensity of price competition t
should be seen as a function of an exogenous institutional variable
T and a vector of other variables Z (which may include N ), in prac-
tice the dominant influence on t in the present context was the
variable T and some of the exogenous elements in Z. In such a set-
ting, it may be possible to ignore any feedback effects from N on t.
An alternative, although not very different, line of defense of the
assumption that t can be taken as exogenous is to say that even if t
were made endogenous, it would be possible to derive and test a
conditional theoretical prediction, provided that t is observable. Sup-
pose, for instance, that we have tðT; ZÞ, where the vector Z may in-
clude N. The free-entry condition can then be written as
pi ðS; N; h; tðT; ZÞÞ b f ; Ei: ð4:3Þ

From (4.3) it is easy to derive a prediction which is conditional on the


known change in t, namely, that a change in the institutional variable

12. According to a stronger version of the efficiency differences approach, the number
of firms need not fall even though concentration should rise following an intensifi-
cation of price competition. In other words, changes in concentration would be mostly
driven by changes in the distribution of market shares rather than by changes in firm
numbers. This hypothesis is discussed in detail in chapter 7. However, it is rejected by
the empirical evidence reported in that chapter.
100 Chapter 4

T must result in a fall in N if and only if t is larger at the new equi-


librium.13 Having established the increase in t in chapter 2, we are
now in a position to test this conditional prediction. In other
words, a reduced-form analysis of the effect of a change in the insti-
tutional variable T on market structure will be sufficient for testing
the theoretical prediction of a positive effect of price competition on
market structure.

4.3 A Specific Example: The Linear Demand Model

I will now briefly illustrate the general mechanism described in the


previous section using a specific theoretical model. This will also
serve as a benchmark, since an extended version of this model will
be used in chapter 5 to illustrate some of the results that apply to
advertising-intensive industries. Consider an exogenous sunk cost
industry producing a potentially infinite number of varieties of a
horizontally differentiated product. Competition in the industry is
described by a two-stage game as follows. There are N0 potential
entrants, each with the capacity to produce a single variety of the
product. At stage 1 they decide whether or not to enter at an exoge-
nously given sunk cost of entry f . At stage 2 those firms that have
entered simultaneously set quantities. We assume that N0 is suffi-
ciently large so that at any equilibrium of the game there is at least
one nonentering firm.
There are S identical consumers, and the utility function of each
consumer takes the form

X XX
U¼ ða1 x i  a2 xi2 Þ  a2 s x i xj þ M: ð4:4Þ
i i j<i

This is a standard quadratic utility function and it has previously


been used, sometimes with small variations, by Spence (1976), Dixit
(1979), Vives (1985), Shaked and Sutton (1990), Sutton (1997b, 1998),

13. Brander and Spencer (1985) examine a model of a homogeneous-good industry


with free entry that allows for a two-way link between the number of firms and a
reduced-form competition measure (in particular, a conjectural variations parameter).
As expected, they find that a shift in competition policy that makes collusion more
difficult for any given number of firms (in my notation, an outward shift of the func-
tion tðT; ZÞ) reduces the equilibrium number of firms.
Price Competition and Concentration 101

and Symeonidis (2002), among others.14 The x i ’s are the quantities


demanded of the different varieties of the product in question, while
P
M ¼ Y  i pi x i denotes expenditure on outside goods. This utility
function implies that the consumer spends only a small part of her
income on the industry’s product (which also ensures that the max-
imization of U has an interior solution), and hence income effects
on the industry under consideration can be ignored and partial equi-
librium analysis can be applied.15 The parameter s, s A ð0; 2Þ, is an
inverse measure of the (exogenous) degree of horizontal product
differentiation: in the limit as s ! 0, the goods become indepen-
dent, while in the limit as s ! 2, they become perfect substitutes. The
parameter s is a basic taste parameter and cannot be influenced by
firms in the industry. It can be seen as an industry-specific measure
of the degree to which demand is diversified among users with dif-
ferent preferences or requirements. Alternatively, it may reflect the
degree of fragmentation of demand caused by transport costs or
trade barriers. Finally, a1 and a2 are positive scale parameters.
The consumer’s inverse demand for variety i is given by

X
pi ¼ a1  2a2 x i  a2 s xj ð4:5Þ
j0i

in the region of quantity spaces where prices are positive. Let N de-
note the number of varieties offered; this is also the number of firms
that have entered the industry. Inverting the system of N equations,
we derive the consumer’s demand function for variety i:
P
ða1  pi Þ½2 þ sðN  2Þ  s ða1  pj Þ
j0i
xi ¼ ð4:6Þ
a2 ð2  sÞ½2 þ sðN  1Þ
in the region of prices where quantities are positive. It can be easily

14. This demand system was introduced by Bowley (1924). A different type of qua-
dratic utility function, which includes the number of firms as a parameter, was pro-
posed by Shapley and Shubik (1969) and was used in slightly different form in Shubik
(1980). See Martin (2002) for a detailed comparison of the two models. Sutton (1997b,
1998) provides a comprehensive analysis of a quantity-setting oligopoly using the
Bowley demand system. This seems a natural choice in the present context, where the
number of firms is endogenous, and hence its inclusion as a parameter in the utility
function seems inappropriate.
15. See Shaked and Sutton (1990) for a discussion of how such a utility function can
also be derived by aggregating the preferences of heterogeneous groups of consumers.
102 Chapter 4

seen that x i is linear and decreasing in pi , and linear and increasing


in pj , Ej 0 i. Since there are S consumers, firm i sells quantity Sx i .
Let each firm have a constant marginal cost of production c, where
c < a1 .
There are two different ways to model the intensity of short-run
‘‘price’’ competition in the present case. The first is standard and
involves comparing the joint monopoly outcome with the Cournot-
Nash equilibrium of the second-stage subgame. The second is an
attempt to allow for imperfect collusion and involves assuming
that at the second-stage subgame each firm maximizes the sum of
its own (gross) profit and a fraction l of the (gross) profit of each of
the remaining firms. The parameter l, 0 a l a 1, is then an inverse
measure of the intensity of short-run competition, with l ¼ 0 corre-
sponding to the Cournot-Nash equilibrium and l ¼ 1 correspond-
ing to perfect collusion. Intermediate values of l represent imperfect
collusion and may be justified by reference to some implicit dynamic
model of collusion, a reduced-form representation of which is the
quantity competition subgame of the present model.16
Of course, despite the fact that l may be a reasonable choice for a
continuous collusion parameter, chiefly because of its properties in
the final-stage subgame, it remains a rather arbitrary way of model-
ing collusion. It should therefore be emphasized that this parameter
is being used here merely to illustrate the results of the previous sec-
tion, which are cast in terms of a continuous competition param-
eter, within the context of the linear demand model. In what follows,

16. The parameter l was first suggested by Edgeworth (1881), who called it the ‘‘co-
efficient of effective sympathy’’ and applied it to the study of individual behavior. It
has been used in oligopoly models by Cyert and deGroot (1973), who named it ‘‘coef-
ficient of cooperation,’’ Kuenne (1980), and Shubik (1980), among others. Note that l is
free from some of the theoretical problems encountered in other approaches to mod-
eling the intensity of short-run competition by way of a reduced-form parameter, such
as the conjectural variations approach.
What also justifies the use of l as a reduced-form competition measure is its prop-
erties in the final-stage subgame: it can be checked that, for given N, the equilibrium
price, price-cost margin, and profit in the second-stage subgame increase and the
equilibrium quantity falls as l rises (the degree of collusion increases). None of the
other exogenous variables that affect gross profit, namely, s or c, have properties sim-
ilar to those of l in the second-stage subgame. Of particular interest in this respect
are the properties of s, since this has often been used as a measure of the intensity of
competition. It can be checked that a fall in s (i.e., an increase in the degree of product
differentiation) increases both the equilibrium price and the equilibrium quantity of the
second-stage subgame.
Price Competition and Concentration 103

l will be treated as a continuous variable, but it will be useful to bear


in mind that all the results on the effect of changes in l on mar-
ket structure are exactly the same if l can take only two values, 0
and 1.
Let me again emphasize that treating the intensity of short-run
competition as an exogenous parameter is a simplification motivated
by the present empirical context. In principle, one could think of l
as a function of an exogenous institutional variable and a vector of
other variables, which may include N.17 The present model would
then be appropriate for a situation where the dominant influence on
l is a change in the institutional variable, since in such a setup it
would be plausible to assume that the feedback effects from N on l
are small (i.e., ql=qN A 0) and can therefore be ignored.
I will now solve for a symmetric subgame perfect equilibrium
in pure strategies in the above two-stage game. Using backward in-
duction, I will start by assuming that N firms have entered at stage 1
of the game and solve for the equilibrium of the second-stage sub-
game. I will then go back to stage 1 to analyze how N is determined,
assuming that firms can anticipate, when making their entry deci-
sions, how the second-stage subgame will be played for any given N.
At stage 2, firms compete by setting quantities. Each firm chooses
P
x i to maximize Pi ¼ pi þ l j0i pj , where
 X 
pi ¼ Sð pi  cÞx i ¼ S a1  c  2a2 x i  a2 s xj x i :
j0i

Solving the system of N symmetric first-order conditions, we obtain


the equilibrium value of x in the second-stage subgame as a function
of the number of firms N:

a1  c
xðNÞ ¼ :
a2 ½4 þ sð1 þ lÞðN  1Þ
The profit of each firm is then given by
Sða1  cÞ 2 ½2 þ slðN  1Þ
pðNÞ ¼ : ð4:7Þ
a2 ½4 þ sð1 þ lÞðN  1Þ 2
It can be easily checked that pðNÞ is increasing in S, decreasing in s,

17. Within this more general formulation, industry A would be defined as being more
collusive than industry B if l A ðNÞ > l B ðNÞ, EN.
104 Chapter 4

and decreasing in N for N > 1. Also, pðNÞ is increasing in l for all


l A ½0; 1Þ and attains its maximum at l ¼ 1. In particular we obtain

qpðNÞ Ssða1  cÞ 2 ½4 þ slð1 þ lÞðN  1Þ


¼ ; ð4:8Þ
qN a2 ½4 þ sð1 þ lÞðN  1Þ 3
which is negative, and
qpðNÞ Ss 2 ða1  cÞ 2 ð1  lÞðN  1Þ 2
¼ ; ð4:9Þ
ql a2 ½4 þ sð1 þ lÞðN  1Þ 3

which is positive for l A ½0; 1Þ and equal to zero for l ¼ 1.


Now at stage 1, the free-entry condition implies that firms enter
the industry as long as the profit they can make in the second-stage
subgame covers the cost of entry. Assuming, for simplicity, that N is
a continuous variable, the free-entry condition implies

pðN  Þ ¼ f ; ð4:10Þ
where N  is the long-run equilibrium number of firms. Given (4.8)
and (4.9), it is then easy to see from the total differential of equation
(4.10) that, for any l A ½0; 1Þ, dN  =dl > 0 (i.e., the equilibrium number
of firms increases as the degree of collusion rises). For l ¼ 1 (i.e.,
at the perfectly collusive point), pðNÞ is maximized and, given (4.8)
and the exogenously fixed f , so is N  . Thus more price competition
raises concentration in an exogenous sunk cost industry. It is also
easy to check that dN  =dS > 0 and dN  =ds < 0, that is, the number of
firms increases with market size and falls as the products become
less differentiated.
It is straightforward to check that the assumption of quantity-
setting firms is innocuous in this model, for similar results are de-
rived for the case of price-setting firms. Moreover, the results do
not depend on the assumption of single-product firms either, because
similar results are obtained when we allow each firm to produce M
varieties, at least for the case where M is exogenously fixed and is
the same across firms.

4.4 The Data

The empirical analysis in this chapter essentially involves a compar-


ison of the evolution of concentration after 1958 between those ex-
ogenous sunk cost industries affected by the 1956 Restrictive Trade
Price Competition and Concentration 105

Practices Act and those not affected—controlling for other factors


that may have influenced concentration during the period examined.
The industry definitions that I will use in this chapter are taken from
the official UK concentration statistics and correspond to the four-
digit level of aggregation.18 The concentration data are available
only for 1958, 1963, 1968, and 1975–1977. Since the legislation did not
have any significant effect before the first Court cases were heard
in 1959, the 1958 observations are all before the natural experiment
took place. Moreover, since competition did not break out imme-
diately in several industries, the impact of the Act was felt at least
until the late 1960s, and sometimes until the early 1970s. Thus
we should expect to see only the short-run effects of competition
by 1963, while by 1968 much of the long-run effect should have been
realized, and by the early 1970s the full long-run effect had certainly
been realized.
This section describes, in considerable detail, the construction of
the data set for this chapter. Since many of the variables used here
will also be used in later chapters, the discussion in this section is of
relevance, and serves as a reference, for data sets used in later chap-
ters as well. Thus the various data sources and the variables used are
described below without any particular focus on one or the other
class of industries. Readers who would prefer to skip the details of
the construction of the data set may go directly to section 4.5, where
a summary description of all the variables used in the econometric
analysis in this chapter is provided.

Concentration

My concentration measure is the share of the five largest producers


in the total sales revenue of UK firms, excluding firms (or plants)
with fewer than twenty-five employees. I will use Census-based data
at the four-digit (or ‘‘product group’’) level of aggregation for 1958,
1963, 1968, and 1975, taken from three official publications: (1) sum-
mary table 5 of the Report on the Census of Production, 1963 (London:
H.M.S.O., 1967); (2) summary table 44 of the Report on the Census of
Production, 1968 (London: H.M.S.O., 1971–1974); and (3) Statistics of

18. Recall that in the system of Census industry definitions, the industry categories
become progressively narrower as digits are added to the industry code. Thus each
three-digit industry includes a number of four-digit industries, each of these includes a
number of five-digit industries, and so on.
106 Chapter 4

Product Concentration of UK Manufactures for 1963, 1968 and 1975,


Business Monitor PO 1006 (London: H.M.S.O., 1979).19
The measurement of concentration presents a number of difficulties
(see Utton 1970). The first of these is the need to use an appropriate
concentration measure. In the present case, the five-firm sales con-
centration ratio is the only available measure at the four-digit in-
dustry level for the period under study.20 The concentration ratio has
been widely used in the literature on the determinants of market
structure (see Curry and George 1983, Sutton 1991), and it is also one
of the measures recommended by Hannah and Kay in their detailed
comparison of several concentration measures on the basis of a
number of axiomatic criteria (Hannah and Kay 1977). For instance,
Hannah and Kay point out that an important advantage of the con-
centration ratio is the fact that it is not very sensitive to the number
of small firms, which affects both the degree of inequality of firm
sizes and the overall number of firms, but is not a key feature of
market structure. This is important in the present context not only
because small firms often do not produce core industry products, but
also because the evidence from the various data sources on compe-
tition suggests that the British cartels did not usually include all
firms in any given industry, and it was often the smaller firms
that were not cartel members; hence the effect of the 1956 Act on
many small firms in cartelized industries may have been relatively

19. The industry categories in the official concentration statistics are defined in accor-
dance to a number of criteria, the most important of which are (1) that they do not
have very low sales, and (2) that they are reasonably homogeneous.
I will also use a few official concentration ratios for 1968 reported in Hart and Clarke
(1980) but not published in the official statistics, and a few concentration ratios for
1975 taken from Statistics of Product Concentration of UK Manufactures for 1975, 1976 and
1977, Business Monitor PO 1006 (London: H.M.S.O., 1980). I will not use concentration
ratios for 1976–1977, even though such data are available from the above source and/
or from a list of unpublished concentration ratios for 1976–1977 compiled by officials
at the Office of Fair Trading and kindly made available to me by David Elliott. My
reason for not using the 1976–1977 data in addition to the 1975 data is to avoid creat-
ing a panel with a very asymmetric intertemporal structure. I could, of course, use the
1977 (or 1976) figures instead of the 1975 figures. This would barely affect the results,
as can be seen by comparing the results reported here with those reported in Symeo-
nidis (2000a). I have chosen to use 1975 instead of 1977 (or 1976) here for two reasons:
first, because this ensures the maximum possible comparability with other data sets
used in this book, and, second, because in a few cases 1976–1977 concentration data
comparable to those for previous years are not available.
20. For most of the period examined in this book, the plant-level data underlying the
Census of Production are not available, because the individual returns to the Census of
Production have, unfortunately, been destroyed.
Price Competition and Concentration 107

weak. Finally, note that the concentration ratio is as appropriate


for testing the present theory as any other standard concentration
measure, since no presumption is made in the theory about any
specific oligopoly model underlying the observed size distribution of
firms.21
Two other common difficulties with the measurement of concen-
tration are the definition of the relevant market and the treatment
of imports. However, these are not very significant problems in the
present case. Consider, first, the definition of the relevant market. A
typical feature of Census-based concentration ratios is the classifica-
tion of industries mainly according to supply rather than demand
characteristics. For instance, glass containers, metal cans, and plastic
containers may be substitutes in demand, but they are classified as
parts of three different Census industries. However, this is probably
an advantage in the present context, because the British cartels typi-
cally operated within industries defined according to their supply
characteristics. Thus, to continue with the previous example, pro-
ducers of metal cans or plastic containers were not parties to the re-
strictive agreement operated among producers of glass containers.
Hence the supply-side classification is appropriate for examining the
impact of the 1956 legislation.
Another typical feature of Census-based concentration ratios, the
allocation of each plant to a particular industry on the basis of the
core products of the plant, does not apply to the present study (ex-
cept for chapter 7). The UK Census of Production allocates each plant
to a ‘‘minimum list heading’’ (MLH) industry, which roughly corre-
sponds to the three-digit level of aggregation. However, the concen-
tration data I will use relate to ‘‘product groups,’’ which I refer to as
four-digit industries, since the level of aggregation is lower than that
of the Census MLH industries and roughly corresponds to the four-
digit level. These data have been constructed on the basis of sales of
products within each product group by all plants employing at least
twenty-five persons, irrespective of the Census MLH industry to
which any particular plant is allocated.

21. Whenever, on the other hand, explicit use is made of the Cournot model of oli-
gopoly, the Herfindahl index may be a more appropriate concentration measure (see
Cowling and Waterson 1976). This is not the case here. In any event, the Herfindahl
index and the concentration ratio are very strongly correlated. Sawyer (1971), who
estimated approximate four-firm concentration ratios and Herfindahl indices for three-
digit British manufacturing industries in 1963, found a correlation coefficient of 0.94
between the two measures.
108 Chapter 4

Next, consider the treatment of imports. Census-based concentra-


tion data typically relate to sales by domestic firms rather than sales
in the domestic market, and the data used in this book are no ex-
ception. To measure concentration in the domestic market, one would
need to subtract exports from and add imports to the sales figures,
and make some assumptions about the market shares of importers
and the fraction of exports accounted for by the largest domestic
firms. Obviously, such an adjustment is subject to serious data limi-
tations, and things are made even more difficult in the present case
by changes in the industrial classification used in the foreign trade
statistics over time and significant differences between this classifi-
cation and the one used in the Census of Production during most of
the period examined in this book.
Fortunately, however, no such adjustment is necessary for the
present study, since the impact of the 1956 Act was usually on com-
petition between UK firms rather than between firms selling in the
UK market. In particular, in most cases the membership of the British
cartels of the 1950s was confined to UK manufacturing firms. On the
other hand, in some cases importers were also parties to the agree-
ments. Even in those cases, however, it is not at all clear that one
would want to adjust the concentration ratio for imports, because
the foreign producers that supplied the British importers were not
subject to the same constraints as the British producers, and hence
it may be preferable simply to focus on the competition effect on con-
centration among British manufacturing firms.22
Of course, the concentration data used here are not entirely free
from problems. A well-known difficulty with the measurement of
concentration in general is the definition of a firm. A firm (or ‘‘en-
terprise’’) is defined in the UK Census of Production as consisting of
all plants under common ownership or control, and control is taken
to mean a 50% holding of equity capital. However, effective control
may arise with a holding of less than 50%. Another difficulty is that
the degree of heterogeneity differs across Census industries, even at
the four-digit level of aggregation: some are relatively homoge-
neous, while others may comprise two or more submarkets which
are largely independent. Nevertheless, these factors should not

22. Moreover, imports were, on the whole, not significant in the 1950s, but increased
during the 1960s and the 1970s. Hence differences across industries in the evolution of
the import-adjusted concentration ratio would be influenced by differences across
industries in the extent of import penetration.
Price Competition and Concentration 109

cause any appreciable statistical bias in a study of market structure


using panel data.
I now turn to a discussion of problems associated with the use of
the published concentration data. There were two main problems.
First, for several industries, data were not available for all years. The
main reasons for this were the progressive increase in the number of
‘‘product groups’’ covered and some radical changes in product
group definitions over time. For example, data for 1958 were avail-
able for some 200 industries, as compared to about 350 for later
years. In such cases, there was little that could be done. Second,
because of changes in industry definitions over time, the figures were
sometimes not comparable over the entire period. In many of these
cases, it was possible to construct comparable figures by adjusting
the ones published.23 On the other hand, whenever the change in the
industry definition was substantial—in particular, whenever there
was a change in industry sales revenue in excess of 20% from one
definition to the other—I either discarded the noncomparable figures
or, in one or two cases, treated the two definitions as two different
industries.24
A few observations were excluded because the high level of gov-
ernment participation or involvement in certain industries made the
figures meaningless for my purposes. This was the case with locomo-
tives and aircraft throughout the period, and also with the industries
affected by the 1967 nationalization of steel (in the latter case only the
observations for 1968 and 1975 were discarded). Some observations
were excluded because the reported concentration ratio was not the

23. In particular, note that summary table 5 of the 1963 Census of Production contains
data for 1958 and 1963, while Statistics of Product Concentration of UK Manufactures for
1963, 1968 and 1975 contains data for 1963, 1968, and 1975. Now noncomparability was
usually the result of a small change in industry definition from one table to the other
rather than from one year to the next within a single table. Hence concentration (and
sales revenue) figures were available for 1963 under both definitions. In such cases,
and provided that the change in industry definition was minor, I applied a simple
proportional adjustment to some of the published figures to ensure comparability over
time. To make sure, in such cases, that the difference in the 1963 figures across tables
was due to a change in industry definition rather than to the revision of a figure, I
checked the individual Census industry reports.
24. Sometimes the noncomparability was due to a change in industry definition from
one year to the next within the same table. Census industry reports were then exam-
ined to assess the magnitude of the change. Usually it was minor (less than 10%);
concentration ratios were then left unchanged—although sales figures were adjusted
on the basis of information from the Census industry reports. Whenever the change
was large, the noncomparable figures were discarded.
110 Chapter 4

five-firm concentration ratio.25 Finally, I excluded a small number of


industries for which concentration data were available only for 1968
and 1975, since these were of little relevance for the study of the
effects of the 1956 legislation.

Market Size

Market size was defined in section 4.3 as a parameter that shifts


quantity demanded at any given price. In practice, however, relative
prices change over time for various reasons, including exogenous
shifts in demand or marginal cost, changes in product quality, and
changes in the product mix within an industry. It may therefore be
useful to think of any increase in the profit opportunities in an in-
dustry, other than one induced by a change in the intensity of price
competition or in endogeneous sunk costs, as an increase in market
size. An ideal proxy for market size should therefore capture varia-
tions in marginal cost, demand, or the product mix over time, but
should be independent of changes in the intensity of price competi-
tion or in sunk costs. For instance, the quantity that would be
demanded if price were equal to marginal cost is one measure that
fits this description, at least for the case where marginal cost is con-
stant and the same across firms.
In practice, there are two possible proxies for market size: sales
revenue deflated by the general producer price index, and sales rev-
enue deflated by an industry-specific producer price index. The for-
mer does not control for changes in relative prices, while the latter
does. It is difficult to say which proxy is better, since this will depend
on the particular form of the demand function.26 Since none of the
two proxies is generally superior to the other and neither of the two
is independent of the intensity of price competition (as an ideal proxy
should be), both will be tried in the empirical model of the next sec-

25. Nearly all of these were cases where the total number of firms in the industry
(excluding those with fewer than twenty-five employees) was slightly larger than five.
In order to avoid the disclosure of information on particular firms, the concentration
statistics then report an n-firm concentration ratio equal to 1 (where n > 5 is the total
number of firms).
26. For example, with a constant-elasticity demand function the total number of profit
opportunities in the market is simply proportional to the value of sales, for any given
degree of collusion; hence, the first proxy may be preferable. With a linear demand
function, however, the second proxy may be preferable, especially when the changes
in relative prices are mainly driven by changes in relative marginal costs rather than
by shifts in demand.
Price Competition and Concentration 111

tion. Using each of these proxies in turn and comparing the results
has one further advantage in the present case: it strengthens the case
for treating sales revenue as an exogenous variable in these regres-
sions (this point will be further discussed in the next section).
Data on the value of manufacturers’ sales at current net producer
prices at the four-digit level of aggregation were obtained mostly
from the same publications as the concentration ratios, and in some
cases from the industry reports of the 1958, 1963 and 1968 Censuses
of Production and the 1975 issues of Business Monitor, PQ series:
Quarterly Statistics. The figures relate to sales by all plants (for 1968
and 1975) or firms (for 1958 and 1963) employing at least twenty-five
persons. The exclusion of very small plants is probably a minor issue,
since these often produce ancillary rather than core industry prod-
ucts. Moreover, a search through the individual industry reports of
the Census of Production did not reveal any serious discontinuities
caused by the switch from excluding very small firms to excluding
very small plants for any industry in my sample.27
A series of general producer price indices was obtained from the
Annual Abstract of Statistics. Producer price indices for a number of
four-digit industries, covering all or part of the period under study,
have been published in the Annual Abstract of Statistics, in various
issues of the Board of Trade Journal and Trade and Industry, in the
Annual Bulletin of Construction Statistics, and in Business Monitor, PQ
series: Quarterly Statistics. For many industries, however, data are not
available or are not available for all years, so price indices at a higher
level of aggregation or of closely related industries have sometimes
had to be used as proxies. An alternative set of price indices can be
constructed on the basis of data on the volume of output often
reported, along with data on sales revenue, in the individual industry
reports of the 1958, 1963, and 1968 Censuses of Production and in
Business Monitors.28
Both these sets of price indices have shortcomings in addition to
the fact that they are often incomplete. Published price indices are
calculated largely from information supplied by firms and trade

27. Such discontinuities were present in only four industries: granite, limestone, sand
and gravel, and ready-mixed concrete. In these industries, firms that employed more
than twenty-five persons and operated one or more small plants each accounted for a
significant fraction of total sales. None of the four industries is included in any of the
samples used in this book, however, because they were all classified as ambiguous
with respect to their state of competition after 1958.
28. See Nicholson and Gupta (1960) for an early application of this method.
112 Chapter 4

associations on prices of products which are thought to be represen-


tative of their respective industries, so they may not always be ac-
curate, especially for earlier years. Also, especially during the 1950s
and the 1960s, these indices were calculated on the basis of list prices
(i.e., trade discounts were not taken into account at a time when they
may have been increasing in many industries, especially those that
had previously been cartelized). Finally, the published price indices
are based on sales in the domestic market only.
Price indices constructed on the basis of data on the volume of
output are subject to more serious problems. These data are usually
reported, along with data on sales revenue, for a number of product
categories within each four-digit industry, so a price index for the
industry can be constructed by using the output volumes of these
product categories as weights. To the extent, however, that each of
these categories consists of subproducts whose prices or price changes
may vary, the overall procedure will result in some measurement
error for the four-digit industry indices. The error will be more seri-
ous, the more differentiated the products within the industry and the
less detailed the breakdown; in fact, some of the data on the volume
of output are probably meaningless. A second problem stems from
the fact that the volume of output is often not reported for all prod-
uct categories within a given four-digit industry, and it is difficult to
say whether the categories covered are representative. On the posi-
tive side, it should be pointed out that since the corresponding data
on sales value are at net producer prices, trade discounts are taken
into account when price indices are computed on the basis of Census
figures on the volume of output; moreover, both domestic sales and
exports are covered.
The series of industry-specific price indices that I use in this book
were constructed after an evaluation of the likely accuracy of the
available figures on a case-by-case basis. For later years, I normally
used the published price indices at the four-digit level whenever they
were available. For earlier years, I used the output-based figures in
some cases, the published price indices in others, and sometimes I
took an average. Often the choice depended on data availability: in
many cases the output data were incomplete or highly aggregated,
while in other cases, published indices at the four-digit level were
not available.29

29. Whenever both these problems were present, the choice usually was between a
published index for a three-digit or a two-digit or a closely related four-digit industry,
Price Competition and Concentration 113

The problems associated with the use of the published sales data
were largely similar to those mentioned above for the concentration
data. Thus data were often not available for all years. On the other
hand, noncomparabilities caused by small changes in industry defi-
nitions were usually dealt with by adjusting some of the published
figures.

Competition

The competition data were fully described in chapter 3. Here I focus


more specifically on the details of the construction of the key com-
petition variable that is used in this chapter and in the rest of the
book. This is a dummy variable, CHANGE, which takes the value 1
for industries with a change in competition regime sometime after
1958 and 0 for industries with no change in regime (including a few
industries where price agreements continued until the 1970s). Recall
that a change in competition regime is defined to correspond to the
cancellation of significant restrictive agreements covering at least
50% of sales revenue in an industry. No change in regime implies
that less than 10% of an industry was in any way affected by the
legislation. Intermediate cases, defined as ambiguous, were excluded.
Note that CHANGE is an industry-specific variable. The econometric
analysis of the competition effect on concentration using CHANGE
throughout this book will involve interacting this variable with year
dummies in order to test whether the time effects on concentration or
any other endogenous variable after 1958 are different for the two
groups of industries in regressions that also control for other factors,
such as market size and setup costs.
Although normally CHANGE takes the value 1 for industries with
a change of competition regime sometime after 1958 and 0 otherwise,
there is also the special case of eight industries that were affected by
the nationalization of steel in 1967. All of these industries had re-
strictive agreements that were abandoned after 1963. Now since the

on the one hand, and an imperfect output-based price index, on the other. I usually
chose the latter approach in the case of sectors where industries are largely hetero-
geneous with respect to costs (e.g., the food and drink sector), and the former for
sectors where a greater degree of homogeneity with respect to costs is present while
output figures are often problematic because of the high degree of product differ-
entiation (e.g., the engineering sector). Sometimes I computed both indices and took an
average.
114 Chapter 4

nationalization of steel made the concentration figures after 1967 ir-


relevant, only the 1958–1963 concentration data for these industries
are included in my data set. But during the 1958–1963 period, the
collusive agreements were still in place in these industries. Conse-
quently, I had to choose between dropping these industries from my
sample and classifying them as industries without a change in com-
petition regime after 1958. I chose to do the latter; in any case, the
results reported below are barely affected by these eight industries.
The main feature of the present approach to modeling the com-
petition effect is that it does not impose any structure on the com-
petition data regarding the timing of the impact of the legislation.
As pointed out above, in several industries competition was slow
to emerge. Moreover, there must have been a time lag between the
emergence of competition and the realization of any effect this may
have had on long-run equilibrium concentration or on other vari-
ables. For these reasons, there is some uncertainty about the timing of
the impact of the Act in each industry. My approach involves letting
the data reveal how the short-run effect of the legislation compares
with the long-run effect, for manufacturing industry as a whole.30

Setup Cost

There are no data for setup costs, so two different proxies were con-
structed. The first of these was constructed by defining a measure of
minimum efficient scale relative to industry size and multiplying it
by the total value of capital stock in the industry (see Sutton 1991).
I used the simplest possible measure of minimum efficient scale,
namely, the size of the average plant. Divided by industry size, this
becomes equal to the inverse of the number of plants in the industry.
Hence the first proxy used for setup cost is the capital stock of the
average plant. A measure of minimum efficient scale based on the

30. An alternative approach to modeling the competition effect was used in Symeoni-
dis (2000a, 2000b). It involved constructing, on the basis of certain plausible assump-
tions about the timing of the impact of the legislation, an industry-year-specific dummy
variable that takes the value 1 for ‘‘collusive’’ and the value 0 for ‘‘competitive’’
industry-year pairs. This alternative approach imposes more structure on the data and
does not allow for an analysis of the short-run impact of the 1956 Act. On the other
hand, it can take into account some information about dates when particular agree-
ments were abandoned. In any case, the two approaches yield very similar results re-
garding the long-run effect of the Act on concentration (see Symeonidis 2000a).
Price Competition and Concentration 115

size distribution of plants might have been more appropriate, but


such a measure could not be used because of data limitations. Even
the data on plant numbers and capital stock are only available at the
three-digit level of aggregation (i.e., for Census MLH industries).
Moreover, the capital stock figures are estimates rather than pri-
mary data. As a result of these conceptual and practical problems,
this proxy is a rather imperfect one. A further difficulty is that
industries differ in the degree to which setup costs are sunk. Ideally,
one would want to relate concentration to that fraction of the setup
cost which is not recoverable on exit (see Kessides 1990), but estimates
of this across industries and over time are not available in the present
context.
These difficulties should not be overemphasized, however. For one
thing, there are, on average, only about two four-digit industries for
each three-digit industry in my sample. More important, if one uses
an empirical model of concentration with industry-specific effects, as
I will be doing in the next section and in the following chapters, one
need not assume that all four-digit industries within any given three-
digit industry are similar with respect to setup cost f , or that the
capital stock of the average plant is an accurate measure of f , or that
all setup costs are sunk costs. All one needs to assume is that the
change in the capital stock of the average plant is roughly similar for
all four-digit industries within any given three-digit industry, that
this change is an adequate measure of the change in f , and that the
degree of ‘‘sunkenness’’ of setup costs is roughly constant for each
industry over time (although it can vary across industries). In par-
ticular, assume that K it ¼ gi KJt , Nit ¼ di NJt , and fit ¼ yi ðK=NÞit , where
K is capital stock, N is the number of plants, and i denotes a four-
digit industry within the three-digit industry J. It then follows that
ln fit ¼ ln yi þ ln gi  ln di þ lnðK=NÞJt . Hence the three-digit lnðK=NÞ
can be used as an explanatory variable, while the term ln yi þ ln gi 
ln di will be part of the four-digit industry-specific effect, provided
that yi , gi , and di are constant over time for each industry i.31
A standard argument against using measures of minimum efficient
scale based on the number or the size distribution of plants in empir-
ical models of concentration is that such measures are, by definition,

31. In fact, the inclusion of time effects in the empirical specification relaxes even more
the assumptions on the parameters yi , gi , and di , since they are also allowed to change
over time, albeit in a uniform way across industries.
116 Chapter 4

correlated with concentration (Davies 1980). To avoid this difficulty,


the capital-labor ratio K=L has sometimes been used as a proxy for
technical economies of scale. Although it may be more difficult to
justify the use of the capital-labor ratio as a proxy for the setup cost
f , it seems plausible that the change in the capital-labor ratio is a
reasonable measure of the change in f and is roughly similar for all
four-digit industries within any given three-digit industry. The
arguments made in the previous paragraph for the three-digit K=N
are therefore also valid for the three-digit K=L.
Capital stock was defined as plant and machinery; buildings and
vehicles were not included because these are to a large extent recov-
erable on exit.32 Estimates of the capital stock (in constant prices) at
the three-digit level of aggregation were taken from O’Mahony and
Oulton (1990). These authors have computed these estimates from
Census investment data using the perpetual inventory method. They
have tried various assumptions about asset lives and depreciation
rates, and have therefore constructed a number of different capital
stock series. Their preferred estimates are the net stock estimates
derived on the assumption of fixed and ‘‘short’’ asset lives and ex-
ponential depreciation rates. I used these figures for 1958, 1963, and
1968; I also computed figures for 1975 on the basis of their 1973 and
1979 estimates, and on the assumption of a constant annual change
in the capital stock between 1973 and 1979.33
Overall, the capital stock estimates of O’Mahony and Oulton can
be treated as reasonably accurate, although one should also be clear
about their limitations. These are caused not only by the fact that
the estimates were constructed on the basis of certain assumptions
that may not hold in a uniform way across industries, but also by
imperfections in the raw data (i.e., the Census investment figures).

32. Part of the cost of plant and machinery may also be recoverable on exit. As pointed
out above, however, differences across industries in the degree to which setup costs
are sunk will be largely captured in panel regressions by the industry effects.
33. A small complication with the 1975 figures has arisen from the fact that the
O’Mahony and Oulton (1990) capital stock estimates for 1979 are not adjusted to con-
trol for the effect of the apparently considerable amount of premature scrapping of
capital assets after 1973. However, the authors provide an estimate for the extent of
premature scrapping for all manufacturing industry after 1973. I therefore used this
estimate to adjust my 1975 capital stock figures, on the assumption that the extent of
premature scrapping was constant for every year between 1973 and 1979 and similar
across industries. Since the relevant period for applying this adjustment in my data set
is very short, the resulting figures for 1975 are very similar to the unadjusted ones.
Price Competition and Concentration 117

There are two main problems with the raw data, both of which are
acknowledged by the authors.
The first problem is that the investment data before 1948 are less
reliable and are available only at a very high level of aggregation,
namely, for eleven industry groups covering the whole of manu-
facturing. The use of these data to construct capital stock figures for
1954 and later years introduces a potentially significant measure-
ment error, especially for 1954 and 1958. On the other hand, for 1963
and later years the influence of the pre-1948 investment on current
capital stock is negligible.
The second limitation relates to the way investment is defined in
the Census of Production. The Census figures are for acquisitions
minus disposals of capital assets, but the definition of disposals
refers only to the selling of assets; it does not include the scrapping
of assets either by firms that exit the industry or by remaining firms.
Of course, O’Mahony and Oulton assume finite asset lives and allow
for depreciation, so ultimately any scrapped assets are eliminated
from the capital stock estimates. But this happens with a significant
lag, and there may be measurement error induced by the fact that
certain industries experience a higher rate of exit or more rapid
technological change (leading to the scrapping of assets) than others
during a certain period.
The potential effect of these measurement errors on the results of
interest in this book is not clear a priori. Fortunately, it may not be
very significant in practice, as is suggested by the similarity of results
using the benchmark estimates of capital stock to results using some-
what more refined estimates of capital stock whenever such more re-
fined estimates could be constructed (see chapter 7, section 7.4, for a
more detailed discussion of this important point).
Data on the number of plants (or Census ‘‘establishments’’) at the
three-digit level of aggregation were taken from the Census of Pro-
duction (various years). There were a number of practical problems
in constructing a series of comparable figures for plant numbers, and
a brief discussion of these problems is now in order.
First, the total number of plants in an industry is greatly affected
by the number of very small plants. These typically account for a
very small fraction of industry capital stock and often produce an-
cillary rather than core industry products. An additional complication
in the present context is that the reported number of plants employing
118 Chapter 4

fewer than eleven persons for all manufacturing industry showed a


50% increase between 1972 and 1975, while the reported number of
plants employing at least eleven persons remained more or less the
same between 1972 and 1975, as well as between 1970 and 1977. One
cannot help but think that the large increase in the number of very
small plants is artificial, whatever the reason. Because of these var-
ious problems with very small plants, I did not take into account
plants with fewer than twenty-five employees in my constructed
figures for plant numbers.
Second, there were some problems with comparability across years.
A number of three-digit industries were substantially redefined be-
tween 1963 and 1968. While fully comparable figures for the number
of plants employing at least twenty-five persons for 1963 and 1968
were published in summary table 1 of the 1968 Census of Produc-
tion, comparable figures for 1958 are not available. Moreover, the
definition of ‘‘establishment’’ was modified between 1968 and 1970,
and this caused an artificial decrease in the reported number of
larger plants between these two years. In many industries the num-
ber fell by up to 20%, and in a few cases by as much as 50%. Clearly,
then, the 1968 and 1970 data are not comparable.
A series of comparable figures for all years in my sample was
constructed as follows. For 1963 and 1968 I used the data reported
in summary table 1 of the 1968 Census of Production. For 1958 I
adjusted the figures reported in the 1958 Census of Production to
ensure comparability with 1963.34 Finally, to construct the figures for

34. Whenever the 1963 industry definition had not been significantly modified in 1968,
I constructed 1958 figures according to the following formula: Estimated N(1958) ¼
½N1 (1958)=N1 (1963)  N(1963), where N1 (1958) is the number of plants with at least
twenty-five employees in 1958, as published in summary table 4 of the 1958 Census;
N1 (1963) is the number of plants with at least twenty-five employees in 1963, as pub-
lished in summary table 4 of the 1963 Census; and N(1963) is the number of plants
with at least twenty-five employees in 1963, as published in summary table 1 of the
1968 Census (the reference table for the series as a whole).
Whenever the 1963 industry definition had been significantly modified in 1968, it was
usually still possible to construct comparable figures for 1958 with the help of more
disaggregated data for 1958 and 1963 published in the individual industry reports of the
1963 Census. These were data on the number of plants operated by firms employing at
least twenty-five employees, and were used on the assumption that the proportional
change in this number over the period 1958–1963 would be equal to the proportional
change in the number of plants employing at least twenty-five persons. I therefore
computed Estimated N(1958) ¼ ½N2 (1958)=N2 (1963)  N(1963), where N2 (1958) and
N2 (1963) are the figures from the 1963 Census industry reports.
Price Competition and Concentration 119

1975 I proceeded as follows. First, I used the proportional change in


the number of employees in plants with at least twenty-five employees
between 1968 and 1970 as a measure of the actual proportional change
in the number of larger plants over the period 1968–1970. Clearly,
this approach is not ideal, but it is certainly much better than using
the reported data on plant numbers for 1968 and 1970 to compute the
1968–1970 proportional change in plant numbers. Since the relevant
period is very small, this should not cause any significant measure-
ment error in the plant number estimates for 1970 and later years.35
Then, to derive the estimated number of plants for 1975, I adjusted
the figures reported in the 1975 volume of Business Monitor, PA series:
Report on the Census of Production on the basis of my estimates for
plant numbers in 1970 (and I also made some further adjustments
because the cutoff points for the plant size distributions changed be-
tween 1970 and 1975).36
Employment data for three-digit industries were taken from sum-
mary table 1 of the 1963 Census, summary table 1 of the 1968 Census,
and summary table 2 of the 1975 Census. Again, because of changes
in industry definitions between 1963 and 1968, comparable data were
sometimes not available for 1958, so they were constructed either
by proportional adjustment of the reported figures or with the help

35. In the large majority of industries, the proportional change in the number of
employees over the period 1968–1970 was small, while the proportional change in the
reported number of plants was larger and always negative, which is consistent with
what was said above about the noncomparability of the 1968 and 1970 figures for plant
numbers. In only a few cases was the reverse true, and I then used the reported figures
for plant numbers, on the assumption that the noncomparability problem was not
serious in these particular industries.
Note that the use of the proportional change in the number of employees as a proxy
for the actual proportional change in the number of plants captures changes in the
number of plants caused by the shutdown of plants or the settingup of new ones, since
these also affect employment. However, measurement error is induced by the fact that
employment in existing plants may also change. But the period 1968–1970 is such a
small part of the overall period examined in this book that this should not make much
difference to the estimated overall change in K/N across industries.
36. More specifically, I started by computing Estimated N(1970) ¼ ½L( 1970)=L( 1968) 
N(1968), where L(1970) and L(1968) are numbers employed in plants with at least
twenty-five employees in 1970 and 1968, respectively, and N(1968) is the number of
plants with at least twenty-five employees in 1968, as published in summary table 1 of
the 1968 Census. Then I computed Estimated N( 1975) ¼ ½N1 (1975)=N1 (1970)  Estimated
N(1970), where N1 (1975) is the number of plants with at least eleven employees in 1975,
as published in the 1975 Census industry reports, and N1 (1970) is the number of plants
with at least eleven employees in 1970, as published in the 1970 Census industry
reports.
120 Chapter 4

of more disaggregated data taken from the 1963 Census industry


reports.37

Defining the Sample of Exogenous Sunk Cost Industries

This chapter focuses on exogenous sunk cost industries (i.e., indus-


tries without significant advertising or R&D expenditures). What
constitutes ‘‘significant’’ advertising or R&D is, of course, somewhat
arbitrary. It is clear that an industry should be classified as an exog-
enous sunk cost industry when its advertising-sales ratio and its
R&D-sales ratio are both relatively low; but the theory is silent as to
what constitutes an appropriate cutoff point. On the other hand, the
1% cutoff point is quite commonly used to classify industries ac-
cording to their advertising or R&D intensity. This is the approach
adopted in this book as well: I will define the group of exogenous
sunk cost industries to consist of all industries with an average or
typical advertising-sales ratio (ADS) of less than 1% and an average
or typical R&D-sales ratio (RDS) of less than 1% over the relevant
period.
Given this definition, it was necessary to determine the average or
typical ADS and RDS for all four-digit industries for which concen-
tration data were available, and then check whether these ratios were
higher or lower than 1%. It should be emphasized that ADS and RDS
are not to be used in this chapter as explanatory variables in the
regressions. They are to be used only to classify industries as exoge-
nous sunk cost, advertising-intensive, or R&D-intensive.

R&D-Sales Ratios

R&D expenditure data for the UK are available for various years
since 1964 at a level of aggregation between the two-digit and the
three-digit (for about thirty to forty ‘‘subsectors’’). They have been
published in Research and Development Expenditure, Studies in Official
Statistics no. 21 (London: H.M.S.O., 1973); Research and Development:

37. In subsequent chapters I also use data on capital stock, the number of plants, and
employment for 1954 and 1973. The capital stock data are from O’Mahony and Oulton
(1990). Employment figures were taken from summary table 1 of the 1958 Census and
from summary table 2 of the 1973 volume of Business Monitor, PA series: Report on the
Census of Production. Data on plant numbers were taken from the industry reports of
the 1958 and the 1973 Censuses. Corrections were made to ensure comparability over
time.
Price Competition and Concentration 121

Expenditure and Employment, Studies in Official Statistics no. 27


(London: H.M.S.O., 1976); and Industrial Research and Development
Expenditure and Employment, Business Monitor MO14 (London:
H.M.S.O., various years since 1972). Data for years before 1964 have
been published in Industrial Research in Manufacturing Industry: 1959–
1960 (London: Federation of British Industries, 1961); Estimates of
Resources Devoted to Scientific and Engineering Research and Develop-
ment in British Manufacturing Industry, 1955 (London: H.M.S.O.,
1958); and Industrial Research and Development Expenditure 1958
(London: H.M.S.O., 1960). The earlier data are somewhat less reli-
able, and even for the 1964–1975 period some of the series are
incomplete, since the number of subsectors varies slightly from year
to year.
A comparison of these various sources suggested that there were
no substantial changes in R&D intensity at the sector level or for
manufacturing as a whole between the late 1950s and the mid-1970s.
Some sectors, such as mechanical engineering, experienced relatively
significant fluctuations in R&D intensity over the period, in contrast
with other sectors, such as chemicals, but the overall changes were not
very large. Hence all the sources were used to classify the industries
according to their R&D intensity. This was measured as the ratio of
company-funded R&D to sales, where R&D includes all current and
capital R&D expenditure by private or public firms, but excludes
royalties and government-financed R&D.38
As already mentioned, however, these data are at a relatively high
level of aggregation. To derive typical R&D-sales ratios for four-digit
industries, the UK data were used to determine average or typical
R&D intensities at the subsector level; in addition, US data were used
as a guide for relative R&D intensities of industries within any UK
subsector. R&D expenditure data for the US, at a level of aggregation
similar to (or sometimes slightly higher than) the UK four-digit in-
dustry level, have been published by the Federal Trade Commission
for the period 1973–1977 in an Annual Line of Business Report for each
one of these years. Note that the direct use of the US data to classify
British industries according to their R&D intensity would not be ap-

38. It is not clear to what extent royalties constitute an endogenous sunk cost. In ad-
dition, there are serious data limitations regarding royalties paid and received by UK
firms at the industry level. On the whole, royalty payments were probably roughly
equal to royalty receipts for UK manufacturing as a whole during the period under
study, although payments were greater than receipts in engineering and the reverse
was true for chemicals.
122 Chapter 4

propriate, since the absolute levels of R&D intensity reported in the


Annual Line of Business reports are, on average, much higher than the
UK R&D-sales ratios for the same period. The assumption made here
is simply that the relative R&D intensities of four-digit industries
within any given subsector will tend to be similar in the two coun-
tries, which is not unreasonable. Thus the US figures were used as a
guide for disaggregating the UK R&D-sales ratios to the four-digit
industry level.39
In fact, this procedure was sometimes qualified to take into ac-
count additional information on R&D intensities in specific British
industries, taken from official or unofficial sources, such as the reports
of the Monopolies Commission, the reports of the National Board for
Prices and Incomes, and case studies contained in Pratten (1971) and
Pavitt (1980).
The overall procedure yielded reasonably accurate R&D-sales
ratios. In any case, these were used only to classify the industries with
respect to R&D intensity, using 1% as a cutoff point. Even if some
industries were misclassified, they must be ‘‘marginal’’ cases with
typical RDS close to 1%, so the empirical results should not be greatly
affected. Finally, there were a few subsectors whose R&D intensity
had changed over time in a way that affected the classification of
industries. These industries were classified on the basis of what
seemed to be their typical status over the relevant period.

Advertising-Sales Ratios

Data on manufacturers’ advertising expenditure come from a number


of sources. Summary table 9 of the 1963 Census of Production and
summary table 4 of the 1968 Census of Production contain data on
advertising expenditure for 1963 and 1968, respectively, at a level of
aggregation somewhat higher than the three-digit level (ninety ‘‘in-
dustry groups’’ are distinguished for 1968, seventy for 1963). Data on
press and TV advertising are also available for all years from market
research sources: the Statistical Review of Press Advertising and the
Statistical Review of Independent TV Advertising, both published by
Legion Information Services, Ltd., until 1963; the Statistical Review of
Press and TV Advertising, published by Legion Information Services,

39. In those cases where the level of aggregation in the Annual Line of Business reports
was higher than the UK four-digit industry level, the assumption was usually made
that all four-digit industries within a particular Line of Business category had the same
R&D intensity.
Price Competition and Concentration 123

Ltd., from 1963 until the early 1970s; and the MEAL Monthly Digest of
Advertising Expenditure, published since 1968 by Media Expenditure
Analysis, Ltd. These data are reported both for individual brands
and for industries, at a relatively low level of aggregation, but they
are often not available for low-advertising industries. Also, they re-
late to the UK market rather than to UK firms. Finally, data on ag-
gregate advertising expenditure by type of advertising have been
published by the Advertising Association in Advertising Expenditure
1960 (London: Advertising Association, 1962) and, subsequently, in
Advertising Quarterly.
A detailed comparison of these various data sources suggested (1)
that the Census figures probably include some expenditures that
represent sales promotion rather than media advertising, and (2) that
the Legion/MEAL figures are, on the one hand, downward-biased
due to incomplete coverage or noncoverage of certain media and, on
the other hand, upward-biased because they are estimated by pricing
the measured amount of advertising according to published rates,
that is, without taking discounts into account (this mainly affects the
estimates for TV advertising).
For my present purposes, advertising expenditure is defined as
media advertising only, since other selling expenses generally affect
marginal cost and are incurred at the price competition stage. The fol-
lowing procedure was therefore adopted for deriving advertising-sales
ratios at the four-digit level for industries covered by Legion/MEAL.
A minimum ADS was derived using Census sales data, approxi-
mately adjusted for net imports according to the Annual Statement
of the Trade of the United Kingdom, and Legion/MEAL advertising
data, adjusted downward to account for discounts from published
rates for media advertising. Changes to this minimum ADS across
time were checked. Moreover, a maximum ADS was estimated using
Census sales and advertising expenditure data. Since these data are
available for industry groups and only for 1963 and 1968, estimates
at the four-digit level were derived by using the Legion/MEAL fig-
ures as a guide for the relative advertising intensities of four-digit
industries within each industry group and by assuming that changes
over time were similar to those of the minimum ADS.
Now as pointed out above, the difference between the two esti-
mates of ADS is due to incomplete coverage or noncoverage of cer-
tain media by Legion/MEAL, and probably also to the inclusion of
some sales promotion expenditures in the Census figures. Although
124 Chapter 4

it is not possible to know the magnitude of these errors in each in-


dustry, one can have an approximate idea of these magnitudes for
manufacturing as a whole by comparing aggregate advertising expen-
diture figures reported in the various data sources, including figures
on aggregate expenditure by type of advertising published by the Ad-
vertising Association. On the basis of that comparison for manu-
facturing as a whole, the average of the two estimates of ADS was
taken as a good approximation to the true ADS in each particular
industry.
Industries for which Legion/MEAL data were not available were
usually easy to classify as having ADS < 1%, but sometimes US data
from the Annual Line of Business reports were used as a guide to
relative advertising intensities of four-digit industries within each
Census industry group.
The advertising-sales ratios derived by this procedure should be
reasonably accurate—and certainly sufficient for the purpose of clas-
sifying the industries with respect to advertising intensity. The only
complication was that, in a few cases, the advertising-sales ratio
changed over time in a way that affected the classification of the in-
dustry. These industries were classified on the basis of what seemed
to be their typical status.

4.5 Empirical Model and Results

The theoretical analysis in this chapter suggests the following em-


pirical model for concentration in exogenous sunk cost industries:
Concentration ¼ CðS; f ; t; wÞ;
where S, f , and t are as defined in section 4.2 and w is a vector of
variables, some of which may be nonmeasurable, industry-specific
characteristics. Note that the vector w includes the degree of hori-
zontal product differentiation, a variable difficult to measure but un-
likely to change significantly for any given industry within a ten-year
or a twenty-year period.
Given that the theoretical predictions are for minimum concentra-
tion, should one estimate a lower bound rather than a regression
line? The estimation of a deterministic lower bound does not allow
for disequilibrium low levels of concentration below the bound, so
it is probably not appropriate in the present context. On the other
hand, estimating a stochastic lower bound by maximum likelihood
Price Competition and Concentration 125

methods is possible only when the least squares residuals are posi-
tively skewed; otherwise the maximum likelihood estimates are
actually the same as the least squares estimates (see Waldman 1982).
As it turns out, the residuals from the least squares dummy variable
models estimated in this section are negatively skewed. This is due
to the fact that some industries experienced large increases in con-
centration during the period examined, so the observations for the
early years have large negative residuals. There may, however, be a
more fundamental problem with the estimation of a lower bound in
the present case, which is due to the panel structure of the data. In
particular, it may not be appropriate to control for industry effects
when estimating a bound. On the other hand, the failure to control
for industry effects would essentially reduce the data set to a pooled
time-series cross section, and it would then be very difficult to iden-
tify a competition effect on concentration because of the prevalence
of industry effects.
The above discussion suggests that the use of standard least squares
regressions is the most appropriate approach in the present case. Note
that an implicit assumption made under this approach is that the the-
oretical predictions for minimum concentration also apply to actual
concentration (i.e., that the mechanism driving the theoretical results
in the case of symmetric single-plant, single-product firms dominates
any other effects that may arise in a more complex setting).

The Sample

The basic sample of industries used in this chapter was constructed


from the set of industries with available concentration data as fol-
lows. First, since the focus of the chapter is on exogenous sunk cost
industries, all industries with average or typical advertising-sales
ratio (ADS) or R&D-sales ratio (RDS) higher than 1% over the rele-
vant period were excluded. Second, I excluded industries with am-
biguous state of competition in 1958 (or, in a few cases, ambiguous
state of competition in the 1960s and early 1970s). Third, I excluded
industries with a switch of competition regime but for which con-
centration data were not available for at least the three core years
1958, 1963 and 1968 (except for one industry without 1958 data,
which I nevertheless included in the basic sample since it abandoned
its restrictive agreement after 1963). However, I did not exclude in-
dustries without a change in regime and with available data for only
126 Chapter 4

Table 4.1
The basic sample of exogenous sunk cost industries
No. of industries with No. of industries without
change of competition change in competition
Data available for regime after 1958 regime after 1958
1958, 1963, 1968, 1975 27 34 (of which 2 collusive)
1958, 1963, 1968 9 3
1963, 1968, 1975 1 15
1958, 1963 0 14 (of which 8 collusive)
1963, 1968 0 4
Total 37 70

Notes: (1) The group of industries without change in competition regime after 1958
includes two industries (glazed tiles and cement) in which collusion continued
throughout the period and for which data are available for all four years.
(2) The group without change in competition regime also includes eight industries
that were affected by the nationalization of steel in 1967. All of these had agreements
that were abandoned after 1963. They have been classified in this group because only
1958–1963 data are used for these industries.
(3) One industry with a change in competition regime and data available for 1963–
1975 is included in the sample, since this industry abandoned its agreement after 1963.

a subset of the period 1958–1968. This left an unbalanced sample


of 107 exogenous sunk cost industries, including 37 with canceled
agreements, and 364 observations. In most cases, observations were
available for all four years in the sample (1958, 1963, 1968, and 1975).
In some cases, however, observations were available for only three
years or for only two years. Table 4.1 gives details on the structure of
the basic sample, distinguishing between industries with a change in
competition regime after 1958 and industries without such a change.
The full set of industries and information on key variables are given
in appendix B (table B1).
As can be seen from table 4.1, the panel is quite unbalanced. This
should not be a problem, but there is also an asymmetry in this
sample between industries affected by the 1956 Act and industries
not affected. In particular, industries without a full set of observa-
tions for at least the core period 1958–1968 are excluded if they are
classified as industries having experienced a change in competition
regime ðCHANGE ¼ 1Þ but not if they are classified as industries not
affected by the 1956 Act ðCHANGE ¼ 0Þ. The idea was to include as
much information as possible in the basic sample while avoiding any
potential bias in the comparison of the short-run and long-run effects
of the legislation. It would also be interesting to check whether
Price Competition and Concentration 127

Table 4.2
Concentration in 1958 and competition regime in exogenous sunk cost industries
Mean C5 in 1958
(st. deviation of C5)
Basic sample
Industries with change of competition regime after 1958 0.514 (0.206)
(n ¼ 36)
Industries without change of regime after 1958, including 0.496 (0.281)
8 industries where collusion was abandoned after 1963
(n ¼ 51)
More balanced sample
Industries with change of competition regime after 1958 0.514 (0.206)
(n ¼ 36)
Industries without change of competition regime after 0.435 (0.277)
1958 (n ¼ 37)
Notes: (1) The figures are based on industries with available data for 1958; n indicates
the number of industries.
(2) The group of industries without a change in competition regime includes two
industries where collusion continued throughout the 1960s and the 1970s.
(3) The group of industries without a change in regime in the basic sample (but not in
the more balanced sample) includes eight industries that were affected by the nation-
alization of steel in 1967. If these industries are excluded, the mean C5 in 1958 for the
group without a change in competition regime in the basic sample is 0.455 (and the
standard deviation is 0.283).

the results are substantially the same when this asymmetry between
the two groups is removed. Thus I have also constructed a second
sample, which includes only industries with a full set of observa-
tions for 1958–1968; in the large majority of cases, the 1975 obser-
vation is also available. This ‘‘more balanced sample’’ contains 73
industries, including 36 with a change in competition regime, and
280 observations.

Descriptive Statistics

Some descriptive statistics on initial levels and changes in concen-


tration are reported in tables 4.2 and 4.3. As pointed out in the pre-
vious section, the five-firm sales concentration ratio, C5, is the only
available measure of concentration at the four-digit industry level for
the period under study. Table 4.2 reports descriptive statistics for C5
in 1958 for industries with and industries without a change in com-
petition regime after 1958. There is little difference between the two
groups in the basic sample: the mean of C5 is 0.514 for 36 industries
128 Chapter 4

Table 4.3
Average change in C5, exogenous sunk cost industries: 1958–1968 and 1958–1975
DC5 1958–1968 DC5 1958–1968 DC5 1958–1975
Industries with 0.129 0.125 0.164
CHANGE ¼ 1 (0.117) (0.123) (0.158)
n ¼ 36 n ¼ 27 n ¼ 27
Industries with 0.087 0.085 0.115
CHANGE ¼ 0 (0.090) (0.090) (0.121)
n ¼ 37 n ¼ 34 n ¼ 34
Notes: The figures in the first column are based on seventy-three industries with
available observations for both 1958 and 1968. The figures in the second and third
columns are based on sixty-one industries with available observations for 1958, 1968
and 1975. The figures in parentheses are standard deviations. n indicates the number
of industries.

with CHANGE ¼ 1 and 0.496 for 51 industries with CHANGE ¼ 0.


Note, however, that the latter group includes some collusive indus-
tries. Most of these are dropped in the more balanced sample, and
although the difference between the two groups increases, it is still
not very large.
The fact that average concentration was somewhat higher in the
group of collusive industries may seem puzzling if one expects price
competition to have a positive effect on concentration. However, the
competition effect on concentration will be very difficult to identify
in a cross section of industries because of the importance of industry-
specific characteristics for concentration. Another factor that blurs
the competition effect on concentration in a cross section is the fact
that the two variables are also negatively related because of a third
variable—capital intensity or the level of setup costs, which has a
positive effect on concentration (see the regression results below) but
also increases the likelihood of collusion, as discussed in chapter 3. In
the more balanced sample of exogenous sunk cost industries, for in-
stance, the mean of 1958 ln K=N for thirty-eight industries with col-
lusive agreements in 1958 is 0.40 (with a standard deviation of
0.89), while it is 1.30 (with a standard deviation of 1.19) for thirty-
five industries without agreements. The respective means (standard
deviations) of ln K=L are 1.11 (0.59) for cartelized industries and 0.57
(0.90) for noncartelized industries. These figures confirm the positive
link between capital intensity and collusion in 1958 obtained in chap-
ter 3 for the manufacturing sector as a whole.
Price Competition and Concentration 129

Table 4.3 presents statistics on the average change in C5 over the


periods 1958–1968 and 1958–1975. The table suggests that price
competition has a significant effect on market structure in exogenous
sunk cost industries. Thus the average change in C5 between 1958
and 1975 in twenty-seven industries affected by the 1956 Act and for
which observations are available for both these years was 16.4 per-
centage points. This compares to 11.5 percentage points for thirty-
four industries not affected by the legislation. A comparison between
industries with a large increase in C5 over the period 1958–1975 and
industries with a large decrease in C5 over the same period is also
very revealing. Obviously, the comparison is limited to the sixty-one
exogenous sunk cost industries with available data for both these
years. Of the five industries with the largest rise in C5, four (bread;
single yarn of cotton, glass fibers, and man-made fibers; woven cloth
of man-made fibers in the loom state; rope, twine, net, and manufac-
tures thereof) are industries with CHANGE ¼ 1, and only one (leather,
undressed) has CHANGE ¼ 0. Conversely, of the five industries with
the largest fall in C5, four (fiberboard packing cases; men’s and boys’
tailored outerwear; fish and marine animal oils; bearings, other than
ball bearings and roller bearings, and bushes) are industries with
CHANGE ¼ 0, and only one (metal windows, metal door frames,
etc.) has CHANGE ¼ 1.
These comparisons, as well as the figures presented in table 4.3,
must be interpreted with some caution for two reasons. First, the use
of the untransformed concentration ratio alone could be somewhat
misleading, to the extent that the change in C5 depends, for purely
statistical factors, on the initial value of C5 (see the discussion on
functional form below). Second, changes in some of the other deter-
minants of concentration were not similar in the two groups over
the relevant periods. To control for these factors as well as for time
effects, I now turn to the econometric analysis of market structure.
As it turns out, the econometric results confirm the picture that
emerges from the descriptive statistics of table 4.3.

Controlling for Industry and Time Effects

The econometric model that I will use is a panel data model with
individual-specific effects. These should control for industry-specific
characteristics that are relative stable, and thus unlikely to change
significantly for any given industry within a ten-year or a twenty-
130 Chapter 4

year period. Three time dummies, for 1963, 1968, and 1975, are also
included among the regressors (the benchmark year is 1958). There
are good reasons for this. The average five-firm concentration ratio
across UK manufacturing industries, which had already been rising
throughout the 1950s, increased by eight percentage points—from
about 55.5% to 63.5%—between 1958 and 1968, then changed very
little between 1968 and the mid-1970s (Hart and Clarke 1980, Hart
1985). Now it was during the 1958–1968 period that the impact of
the 1956 Act on competition was mostly realized. If the rise in con-
centration was partly caused by factors other than the 1956 Act or
changes in setup costs and market size, then the estimated coef-
ficients would be seriously biased if the model were specified with-
out time dummies.
Several factors have been cited as having contributed to the rise in
concentration during the 1960s. They include changes in the tax sys-
tem in the mid-1960s that are thought to have encouraged mergers;
economies of scale in product development, distribution, and the
raising of finance; improvements in the system of transport; the
emergence of large retailers with significant bargaining power; and
the progressive opening of the British economy. It is very difficult to
measure these factors at the industry level, and no such attempt will
be made here. However, there is no reason to expect that they would
affect the group of industries with a change in competition regime
and the control group of industries in different ways. To the extent
that their effect has been more or less realized across all industries, it
should be largely captured by the time dummies.

The Effect of Foreign Competition

Of all these factors, the one whose omission from my empirical speci-
fication is the most regrettable is the intensification of foreign compe-
tition caused by the gradual opening of the British economy during
the 1960s and the 1970s. Unfortunately, it is difficult to control for this
in a more satisfactory way. Ideally, one would need some measure of
the extent of foreign competition for each industry across time. Two
possible candidates are the import penetration ratio and the rate of
effective protection. However, there are serious problems, theoretical
and practical, with both of these measures. Estimates of effective
rates of protection are available at a high level of aggregation and for
only some years in my sample; also, they are often subject to mea-
Price Competition and Concentration 131

surement error. The import penetration ratio, on the other hand, is a


poor proxy for the extent of foreign competition: it cannot capture
the effect of the mere threat of competitive imports, it does not take
into account imports by domestic producers (which may not be in
competition with domestic products), and it is itself clearly endoge-
nous. Moreover, the industrial classification used in the foreign trade
statistics during the period examined in this book has been subject to
changes over time and is often difficult to match with the industrial
classification used in the Census of Production.
It should be emphasized, however, that there is no reason to think
that foreign competition may have had a differential effect across the
two groups of industries (i.e., the group with CHANGE ¼ 1 and the
one with CHANGE ¼ 0) after 1958. As pointed out in section 3.4 of
chapter 3, a rough test of the link between the extent of foreign
competition and the incidence of collusion in the 1950s revealed a
negative association, but this was not statistically significant when
controlling for sector effects (see also Symeonidis 1999b). Thus there
is no strong evidence of any difference in initial conditions between
the two groups with respect to foreign competition.
Moreover, although tariff reductions occurred throughout the 1960s
and the 1970s, they became more pronounced after 1967, when the
Kennedy Round was completed (see Morgan and Martin 1975).40
This may be part of the reason why Kitchin (1976) was not able to
find any overall pattern of falling or rising effective protection be-
tween 1963 and 1968 in UK manufacturing. On the other hand, the
effect of the 1956 restrictive practices legislation was mostly realized
between 1958 and 1968 (i.e., before the first stage of the Kennedy
Round tariff cuts).
Finally, there is no evidence that changes in the level of effective
protection were any different between industries with a change in
competition regime and industries in the control group, at least be-
tween 1963 and 1968. Kitchin provides estimates of effective pro-
tection for both these years at a level of aggregation between the
two-digit and the three-digit industry level. Although these esti-
mates are subject to serious limitations, they are sufficient for a rough
comparison. Effective tariff protection increased, according to these
figures, in six out of twelve industry groups that I could classify as
having experienced a change in competition regime, and decreased

40. In contrast, the Dillon Round tariff cuts of 1961 were small and were partially off-
set by the import surcharge of 1964–1966.
132 Chapter 4

in the other six. For industry groups that I could classify as having
experienced no change in competition regime, the respective num-
bers were eight and ten.
In summary, then, it is not unreasonable to argue that the esti-
mated effect of the 1956 Act in this and in subsequent chapters is not
biased by the failure to control for foreign competition.

The Econometric Specification

Since there was no obvious choice of functional form, a number of


specifications were tried. An argument often made in empirical studies
of concentration is that since the concentration ratio is bounded be-
tween 0 and 1, the change in the concentration ratio may depend, for
purely statistical reasons, on its initial value. For instance, the con-
centration ratio cannot increase by much in absolute terms when it is
already very high, and it cannot decrease by much when it is already
very low. Failure to take this into account could result in hetero-
skedasticity, and it could even bias the regression coefficients if there
is a systematic association between the initial value of concentration
in the sample and any of the regressors. In the present case, there
may be a weak association, linear or nonlinear, between the value of
C5 in 1958 and the variable CHANGE (see chapter 3 and table 4.2). To
overcome any potential problems from this, it seemed appropriate to
complement the results using C5 with results using a transformation
of C5 which is not bounded (see Wright 1978). The logistic transfor-
mation, logit C5 ¼ ln½C5=ð1  C5Þ, was therefore tried as an alterna-
tive to the untransformed concentration ratio.
Plots of residuals against fitted values and formal tests for hetero-
skedasticity such as the Koenker test (which is robust to nonnormality
of the residuals) and the Breusch-Pagan test were used to compare the
two specifications. It turned out that heteroskedasticity was present
in both specifications and that, contrary to what is usually expected,
it was not more pronounced in regressions using C5 than in those
using logit C5. Results for both C5 and logit C5 will be reported below.
Preliminary regressions were also performed to compare different
functional forms for the independent variables. As expected, specifi-
cations with log transformations of the exogenous variables were
generally preferable on the basis of the properties of the residuals
and/or tests for nonnested models.
Price Competition and Concentration 133

The above discussion suggests the following specification:


Concit ¼ a i þ b 1 ln Salesit þ b2 lnðK=xÞit þ b3 Y63 þ b 4 Y68 þ b5 Y75
þ b 6 CHANGE  Y63 þ b7 CHANGE  Y68

þ b 8 CHANGE  Y75 þ uit ;


where i stands for industries and t for time periods, and a i is the
industry-specific effect. ‘‘Conc’’ is either the four-digit industry five-
firm sales concentration ratio, C5, or its logistic transformation,
logit C5. ‘‘Sales’’ is the four-digit industry sales revenue deflated
either by the general producer price index ðSSÞ or by an industry-
specific producer price index ðDSÞ; both of these are measures of mar-
ket size. ‘‘K/x’’ is either the three-digit industry capital stock of the
average plant, K=N, or the three-digit industry capital-labor ratio,
K=L; both of these are proxies for setup costs.41 Y63, Y68, and Y75
are time dummies for 1963, 1968, and 1975, respectively.
Finally, the interaction terms should capture any differences in
the evolution of concentration after 1958 between industries with a
change in competition regime ðCHANGE ¼ 1Þ and industries without
such a change ðCHANGE ¼ 0Þ. Thus the coefficient on CHANGE  Y63
measures the effect of the 1956 Act on C5 or logit C5 between 1958
and 1963, the coefficient on CHANGE  Y68 measures the effect be-
tween 1958 and 1968, and the coefficient on CHANGE  Y75 mea-
sures the effect between 1958 and 1975. Obviously, CHANGE is
defined here according to the industry categories used for C5. As
pointed out in the previous section, an attractive feature of the pres-
ent approach to modeling the competition effect is that it does not
impose any structure on the competition data regarding the timing
of the impact of the legislation. Rather, it allows the data to reveal
how the short-run effect of the legislation compares to the long-run
effect for manufacturing industry as a whole.
I should also point out that it is not possible to use a dynamic
panel data model (i.e., include lagged values of C5 or logit C5 among
the explanatory variables) in the present context. Such an approach
would be motivated by the fact that concentration may adjust grad-

41. Recall that since the model includes industry-specific effects, one need not assume
that the three-digit industry K/L or K/N is an accurate measure of setup cost at the
four-digit industry level. All that is required is that the change in K/L or K/N is roughly
similar for all four-digit industries within any given three-digit industry, and is an
accurate measure of the change in setup cost.
134 Chapter 4

ually to its long-run equilibrium level. However, the estimation of


a dynamic model involves taking first differences and then using
lagged values of the endogenous variable as instruments; to apply
this method, one effectively ‘‘loses’’ the first one or two periods in the
panel. In the present case, this would amount to losing much of the
crucial information captured by the interactions of CHANGE with
the time dummies.
In any case, it is not clear that a dynamic specification would be a
significant improvement over the one used here. For one thing, the
years in the panel that I use in this chapter, and also in the panels
used in subsequent chapters, are separated by long periods, typically
five years or more. Although concentration may take longer than five
years to fully adjust to its long-run equilibrium level, much of the
adjustment should occur within a few years. Moreover, my econo-
metric specification does not impose any structure on the competi-
tion data regarding the timing of the impact of the legislation. In
other words, it allows the competition effect on concentration to op-
erate with an arbitrary lag. Of course, this flexibility of the specifi-
cation does not extend to the other determinants of concentration
included in the model, namely, market size and setup costs. Changes
in setup costs, in particular, may affect market structure relatively
slowly, as firms gradually renew their capital stock. However, recall
that setup costs are measured somewhat imprecisely throughout this
book. As a result, part of the effect of changes in setup costs is likely
to be captured by the time dummies in most regressions. I will return
to this issue below, when discussing the regression results. The point
I wish to make here is that (1) the use of a weak proxy for setup cost,
(2) the intertemporal structure of the panel, and (3) the way in which
the competition effect is modeled reduce any potential benefit that
might be derived from specifying a dynamic model for concentration
in the present context.42
A stronger objection to the model used here is that some of the
independent variables may be endogenous. This is probably not a
serious problem for the market size and setup cost proxies, since the
variation in these empirical measures across industries and five-year
periods must be mainly driven by the variation in the corresponding

42. It could also be argued that if lagged concentration were indeed an important
explanatory variable in the model, its omission would show as significant serial
correlation in the residuals. But there is no such evidence in the models estimated
below.
Price Competition and Concentration 135

theoretical variables. In the case of market size, an indirect check of


this claim is provided by a comparison of results using ln DS with
those using ln SS. For example, consider the argument that changes
in sales revenue may be endogenous with respect to changes in the
competition regime since an intensification of price competition may
raise deflated sales revenue.43 If that were indeed a serious problem
in the data, one would expect the results from regressions using
ln DS and those from regressions using ln SS to differ, especially re-
garding the coefficients on the competition variables. As will be seen
below, the results are in fact very similar.
A more serious objection is the potential endogeneity of the vari-
able CHANGE. This is an important issue and needs to be discussed
in detail.

The Key Identifying Assumption

The principal objection regarding the use of the variable CHANGE is


that whatever difference one may observe in the evolution of con-
centration after 1958 between industries with CHANGE ¼ 1 and in-
dustries with CHANGE ¼ 0 may be to some extent due to unobserved
characteristics that differ between the two groups of industries rather
than to the 1956 legislation. If this were indeed the case, then the key
identifying assumption of this book would no longer be valid (see
chapter 1). Unfortunately, it is not possible to test formally for ex-
ogeneity because there are no appropriate instruments for CHANGE.
Those variables that were shown in chapter 3 to be strongly corre-
lated with the incidence of collusion in 1958 cannot be used as in-
struments: capital intensity is also a key determinant of concentration
and is therefore included as an independent variable in the model,
while advertising intensity is not relevant for a sample of exogenous
sunk cost industries and is itself endogenous.
However, there are several reasons to believe that the potential
endogeneity of CHANGE is not a serious problem in the present

43. In fact, it is by no means clear that the breakdown of collusion will lead to a rise in
industry sales revenue when the deflator used is the general producer price index for
all manufacturing. Whether sales revenue increases or decreases will depend on the
elasticity of demand. Although a joint monopoly would operate on the elastic part of
the demand curve, its abolition may well move the industry to the inelastic part of the
demand curve, so sales revenue may rise or fall relative to its original level. Moreover,
collusion need not involve joint profit maximization, so it is not even clear that a col-
lusive industry will be operating on the elastic part of the demand curve.
136 Chapter 4

context.44 First, table 4.2 suggests that the difference in market


structure conditions in 1958 between industries affected by the leg-
islation and those not affected is not very pronounced. This is con-
sistent with the evidence presented in chapter 3, according to which
there was no systematic link between concentration and collusion
across manufacturing industries in 1958. Admittedly, the relative
similarity of initial market structure conditions is not a very decisive
argument against the potential endogeneity of CHANGE in the pres-
ent case, because the two groups of industries were operating under
different conditions in 1958. Indeed, if one looked at the end of the
period under study (i.e., the year 1975), one would find that the
two groups had quite different average concentration ratios then. On
the other hand, the relative similarity of initial market structure
should at least help dispel the concern that some of the difference
that we observe between the two groups after 1958, especially in
regressions using C5, is due to a purely statistical factor, the fact
that the change in C5 may generally depend on the initial value of
C5.
Second, one could argue that even if CHANGE is influenced by
certain variables that also affect concentration and are not included
in the model, these variables are more likely to be part of the industry-
specific effect than of the error term, since the large majority of in-
dustries classified as collusive in 1958 were subject to restrictive
agreements for many years before the introduction of the 1956 legis-
lation. Such correlations between the industry-specific effects and the
endogenous variables, if they exist, will not cause any econometric
difficulties, provided that an appropriate specification is used.
Finally, a powerful indirect check of the claim that the potential
endogeneity of CHANGE is probably not a serious problem would be
to examine the evolution of market structure in the two groups of
industries before 1958. If the two groups were found to be similar in
this respect, then it would probably be safe to conclude that any
observed difference between them in later periods is due to the leg-
islation. Unfortunately, the available concentration data before 1958
are not comparable with the data used in this book because of changes
in industry definitions and the concentration measures used. How-

44. Arguments similar to those that I use below, and in subsequent chapters, to defend
the key identifying assumption in my analysis of the effects of the 1956 Act have also
been put forward in several other recent studies of natural experiments. See, for in-
stance, Chevalier (1995), Eissa (1995), and Eissa and Liebman (1996).
Price Competition and Concentration 137

ever, there are comparable figures for firm numbers in both 1954 and
1958 for a large sample of industries. As will be pointed out in chapter
7, where these data are analyzed in detail, there was almost no differ-
ence whatsoever in the evolution of firm numbers during the period
1954–1958 between the two groups of industries.45
Further evidence can be obtained from a comparison of three-firm
employment concentration ratios for three-digit industries in 1951
and 1958. The data are contained in Armstrong and Silberston (1965),
and are taken from Evely and Little (1960) and the 1958 Census of
Production. Because of changes in industry definitions and other
noncomparabilities between the two years, the sample is small. The
average change in the three-firm concentration ratio C3 during the
period 1951–1958 was 2.5 percentage points for fifteen exogenous
sunk cost industries with a change of competition regime after 1958,
and 2.0 percentage points for eleven exogenous sunk cost industries
without such a change. A test of independence of the respective
means of the change in C3 in the two groups could not reject the null
hypothesis of independence at the 5% level or, indeed, at the 10% or
20% level. Although this sample is very small and the three-digit
level of aggregation is far from ideal, there is clearly no evidence of
any differential underlying trend between the two groups.

Econometric Results

The choice of estimation procedure in panel data models depends on


whether the individual effects are correlated with the independent
variables (Hsiao 1986). If they are not correlated, then a random-
effects specification is preferable, since it produces efficient estimates
by taking account of the unobserved heterogeneity both between
groups and within groups. If the individual effects are correlated
with the independent variables, then the random-effects model pro-
duces inconsistent estimates, so a fixed-effects model must be used
instead. This latter model takes into account only the within-group
heterogeneity, and therefore produces estimates that are less efficient,
although they are always consistent. In the absence of any prior indi-
cation of potential correlation between the individual effects and the

45. A four-year period is, of course, rather short. However, the period 1958–1963
is also rather short, and this does not prevent the difference between the two groups
(presumably caused by the 1956 Act) from emerging sharply during that period. See
chapter 7 for details.
138 Chapter 4

independent variables, a formal test, such as the Hausman test, can


be used to choose the preferred model.
In the present case, the Hausman test has given somewhat mixed
results, largely depending on what proxy is being used for setup
cost: in most regressions using ln K=N the test favors the fixed-effects
model, while in most regressions with ln K=L it favors the random-
effects model. The two sets of results are very similar with respect
to the market size variable and the competition terms, but differ re-
garding the respective explanatory power of the time dummies and
the setup cost proxies. In any case, results from both models are given
below for the basic sample: table 4.4 presents results for the fixed-
effects specification, and table 4.5 reports results for the random-
effects model, using generalized least squares estimation. Results are
reported in table 4.6 for a more balanced sample that includes only
industries with a full set of observations for 1958, 1963, and 1968.
Only results for the random-effects model are presented for the more
balanced sample, because the Hausman test now more clearly fails to
reject this model. The reported standard errors are heteroskedasticity-
consistent, adjusted for small sample bias following MacKinnon and
White (1985).46
The results from all three tables tell a similar story about the effect
of the 1956 Act on concentration in exogenous sunk cost industries.
The coefficients on CHANGE  Y68 and CHANGE  Y75 are both
positive and statistically significant at the 1% or the 5% level in all
regressions. The magnitude of these coefficients implies that the in-
tensification of price competition following the 1956 legislation
raised, on average, the five-firm concentration ratio by about six to
seven percentage points in exogenous sunk cost industries. The effect
was almost fully realized by 1968. Note that this may understate the
impact of price competition on concentration to the extent that there
is measurement error in the construction of CHANGE due to ineffec-
tive or unregistered agreements. Moreover, the coefficient on
CHANGE  Y63 indicates that the effect was only partly realized by
1963, presumably because in several industries competition had not
yet emerged or its effect on concentration had not yet been fully
realized. All these results are not much affected by the choice of de-

46. The reported results are all for models without serially correlated disturbances. I
also estimated a model with an AR1 error structure, and the results were very similar to
those reported here. The estimated coefficient of serial correlation in these regressions
was very low (and sometimes even negative): between 0.05 and –0.12 when using the
basic sample, and between 0.05 and –0.03 when using the more balanced sample.
Price Competition and Concentration 139

pendent variable, C5 or logit C5, or the choice of proxies for market


size and setup cost.
Three other features of the results are worth emphasizing. First,
the effect of market size on concentration is everywhere negative and
statistically significant at the 1% level, which is consistent with the
theoretical predictions of sections 4.2 and 4.3. Moreover, the coeffi-
cients on sales revenue are very similar in these regressions, irrespec-
tive of whether one uses ln SS or ln DS as a proxy for market size, thus
strengthening the case for treating sales revenue as an exogenous
variable.
Second, the use of rather weak proxies for setup cost in these
regressions results in significant correlation between these variables
and the time dummies. In regressions using ln K=N and in random-
effects regressions using ln K=L, the time dummies are not statisti-
cally significant, but the setup cost proxies are highly significant and
have large coefficients. The reverse is true for fixed-effects regres-
sions using ln K=L. Clearly, the impact of the increase in setup costs
across industries during the 1960s and the 1970s cannot be very
clearly separated from the impact of other factors that are probably
captured by the time dummies in these data. This is a fairly general
feature of the results reported in this book, but, as can be seen from a
comparison across tables in this chapter, it has no significant impli-
cations for the results of interest (i.e., those regarding the competi-
tion effect and the market size effect).
Third, two different R 2 s are reported for each regression in table
4.4. The first is derived from transforming the data to obtain devia-
tions from industry means and applying OLS to the transformed
data. The second (denoted R 2LSDV ) is obtained from applying OLS to
the untransformed data after including a set of industry dummies
among the regressors. Thus the difference between the two R 2 s is a
measure of the explanatory power of the industry effects in these
regressions. As can be seen in table 4.4, the industry effects explain
about half the variation in the concentration ratio in these data.

4.6 Concluding Remarks

Competition policy on restrictive agreements in the UK has been an


important factor in increasing concentration in exogenous sunk cost
manufacturing industries. The empirical results of this chapter are
consistent with theoretical models that emphasize the endogeneity of
market structure and the effect of firm conduct on structure. A key
Table 4.4

140
Regression results for concentration in exogenous sunk cost industries (fixed-effects estimation)
Dependent variable: C5 Dependent variable: logit C5
ln SS 0.078 — 0.082 — 0.30 — 0.31 —
(0.016) (0.017) (0.10) (0.10)
ln DS — 0.073 — 0.073 — 0.31 — 0.31
(0.015) (0.016) (0.09) (0.10)
ln K=N 0.089 0.100 — — 0.45 0.49 — —
(0.023) (0.023) (0.14) (0.13)
ln K=L — — 0.022 0.035 — — 0.19 0.23
(0.038) (0.038) (0.21) (0.20)
Y63 0.006 0.004 0.027 0.026 0.07 0.07 0.16 0.16
(0.012) (0.013) (0.014) (0.014) (0.08) (0.08) (0.08) (0.08)
Y68 0.038 0.035 0.081 0.078 0.21 0.21 0.39 0.39
(0.020) (0.020) (0.025) (0.025) (0.11) (0.11) (0.14) (0.13)
Y75 0.038 0.033 0.100 0.093 0.23 0.22 0.48 0.47
(0.027) (0.027) (0.038) (0.039) (0.16) (0.16) (0.21) (0.21)
CHANGE  Y63 0.041 0.040 0.038 0.036 0.09 0.09 0.08 0.07
(0.020) (0.020) (0.021) (0.021) (0.12) (0.12) (0.12) (0.12)
CHANGE  Y68 0.062 0.056 0.057 0.050 0.26 0.24 0.24 0.22
(0.021) (0.021) (0.021) (0.021) (0.12) (0.12) (0.12) (0.12)

Chapter 4
CHANGE  Y75 0.073 0.062 0.067 0.056 0.32 0.27 0.30 0.25
(0.026) (0.026) (0.027) (0.027) (0.14) (0.14) (0.15) (0.15)
R2 0.52 0.51 0.50 0.48 0.45 0.46 0.44 0.44

Price Competition and Concentration


2
RLSDV 0.96 0.96 0.95 0.95 0.95 0.95 0.95 0.95
No. of industries 107 107 107 107 107 107 107 107
No. of industries 37 37 37 37 37 37 37 37
with CHANGE ¼ 1
No. of observations 364 364 364 364 364 364 364 364
2
Notes: Heteroskedasticity-consistent standard errors in parentheses. R is derived from transforming the data to obtain deviations from in-
2
dustry means and applying OLS to the transformed data. RLSDV is obtained from applying OLS to the untransformed data after including a set
of industry dummies among the regressors.

141
Table 4.5

142
Regression results for concentration in exogenous sunk cost industries (random-effects estimation)
Dependent variable: C5 Dependent variable: logit C5
ln SS 0.085 — 0.076 — 0.38 — 0.33 —
(0.012) (0.013) (0.07) (0.07)
ln DS — 0.081 — 0.071 — 0.38 — 0.33
(0.012) (0.013) (0.07) (0.08)
ln K=N 0.123 0.124 — — 0.60 0.61 — —
(0.010) (0.010) (0.06) (0.06)
ln K=L — — 0.140 0.141 — — 0.72 0.72
(0.015) (0.015) (0.08) (0.08)
Y63 0.002 0.001 0.002 0.003 0.04 0.05 0.05 0.06
(0.012) (0.012) (0.013) (0.013) (0.07) (0.07) (0.07) (0.08)
Y68 0.020 0.024 0.019 0.022 0.14 0.16 0.12 0.14
(0.013) (0.013) (0.016) (0.016) (0.07) (0.07) (0.08) (0.08)
Y75 0.008 0.013 0.006 0.002 0.10 0.13 0.002 0.02
(0.016) (0.016) (0.020) (0.021) (0.09) (0.09) (0.11) (0.11)
CHANGE  Y63 0.040 0.040 0.041 0.040 0.09 0.09 0.09 0.09
(0.018) (0.018) (0.018) (0.019) (0.10) (0.10) (0.10) (0.10)
CHANGE  Y68 0.062 0.055 0.067 0.060 0.26 0.23 0.29 0.26
(0.018) (0.018) (0.020) (0.019) (0.11) (0.11) (0.11) (0.12)

Chapter 4
CHANGE  Y75 0.076 0.064 0.085 0.074 0.34 0.28 0.39 0.34
(0.022) (0.022) (0.023) (0.023) (0.13) (0.13) (0.13) (0.13)
Constant 1.486 1.443 1.170 1.119 4.46 4.49 2.91 2.89

Price Competition and Concentration


(0.129) (0.133) (0.133) (0.138) (0.76) (0.79) (0.77) (0.80)
R2 0.51 0.51 0.47 0.47 0.45 0.45 0.42 0.42
Hausman statistic 2.80 0.71 22.29 17.88 3.86 1.45 15.99 12.91
Prob-value 0.95 0.99 0.004 0.02 0.87 0.99 0.04 0.11
No. of industries 107 107 107 107 107 107 107 107
No. of industries 37 37 37 37 37 37 37 37
with CHANGE ¼ 1
No. of observations 364 364 364 364 364 364 364 364
Note: Heteroskedasticity-consistent standard errors in parentheses.

143
Table 4.6

144
Regression results for concentration in exogenous sunk cost industries: more balanced sample (random-effects estimation)
Dependent variable: C5 Dependent variable: logit C5
ln SS 0.084 — 0.076 — 0.37 — 0.33 —
(0.014) (0.014) (0.07) (0.08)
ln DS — 0.078 — 0.067 — 0.37 — 0.32
(0.014) (0.015) (0.08) (0.08)
ln K=N 0.122 0.126 — — 0.61 0.62 — —
(0.013) (0.013) (0.07) (0.07)
ln K=L — — 0.124 0.127 — — 0.68 0.68
(0.021) (0.021) (0.11) (0.11)
Y63 0.003 0.003 0.003 0.002 0.01 0.01 0.02 0.02
(0.014) (0.014) (0.016) (0.016) (0.08) (0.08) (0.09) (0.09)
Y68 0.024 0.027 0.033 0.035 0.15 0.16 0.16 0.17
(0.016) (0.016) (0.019) (0.020) (0.08) (0.08) (0.10) (0.10)
Y75 0.011 0.013 0.012 0.014 0.11 0.12 0.06 0.07
(0.019) (0.020) (0.025) (0.026) (0.10) (0.10) (0.13) (0.13)
CHANGE  Y63 0.042 0.042 0.044 0.044 0.12 0.12 0.13 0.13
(0.020) (0.020) (0.020) (0.021) (0.11) (0.11) (0.11) (0.11)
CHANGE  Y68 0.059 0.051 0.062 0.055 0.26 0.23 0.28 0.25
(0.020) (0.019) (0.021) (0.021) (0.11) (0.11) (0.12) (0.12)

Chapter 4
CHANGE  Y75 0.074 0.062 0.080 0.069 0.34 0.28 0.37 0.32
(0.023) (0.024) (0.024) (0.025) (0.14) (0.14) (0.14) (0.14)
Constant 1.469 1.403 1.175 1.080 4.38 4.34 2.87 2.70

Price Competition and Concentration


(0.149) (0.154) (0.152) (0.157) (0.82) (0.88) (0.84) (0.88)
R2 0.55 0.54 0.51 0.50 0.49 0.49 0.46 0.46
Hausman statistic 1.87 2.91 15.53 12.88 4.81 2.48 12.07 10.43
Prob-value 0.98 0.94 0.05 0.12 0.78 0.96 0.15 0.24
No. of industries 73 73 73 73 73 73 73 73
No. of industries 36 36 36 36 36 36 36 36
with CHANGE ¼ 1
No. of observations 280 280 280 280 280 280 280 280
Note: Heteroskedasticity-consistent standard errors in parentheses.

145
146 Chapter 4

prediction of these models is that in exogenous sunk cost industries,


an intensification of price competition, brought about by institutional
changes such as the introduction of cartel laws or a higher degree of
economic integration, will cause a rise in concentration. This basic
theoretical insight gets strong support from the evidence presented
here. In particular, the results indicate that, on average, cartel policy
in the UK raised the five-firm concentration ratio by about six to
seven percentage points between 1968 and 1975 in exogenous sunk
cost industries. To put this figure in some perspective, note that
nearly half of the entire British manufacturing sector was cartelized
in the 1950s and that the average five-firm concentration ratio across
all manufacturing industries increased by about eight percentage
points between 1958 and 1975.
Moreover, the view taken in this chapter is that the increase in
concentration in previously collusive British industries during the
1960s was the inevitable result of a general structural mechanism. In
other words, the initial level of concentration was simply not sus-
tainable once competition intensified, so market structure had to
adjust to restore profits in the long run. Additional insight on this
mechanism will be provided in chapter 7, which will examine the
effect of the 1956 restrictive practices legislation on profitability in
the short run and in the long run.
The mechanism driving the theoretical results of this chapter—
namely, the requirement that market structure adjusts under free
entry and exit to satisfy the requirement that gross profits cover sunk
costs—is fully compatible with the view, expressed in Swann et al.
(1974, pp. 176–177) and Curry and George (1983), among others, that
a toughening of cartel policy would cause firms to try to restore
market power through mergers and would thus lead to higher con-
centration. This view places less emphasis on structural effects than
on the pursuit of market power that is assumed to underlie firm
behavior in oligopolistic industries. The key difference between a
purely behavioral interpretation of the effects of the 1956 legislation
and an interpretation that emphasizes the operation of a structural
mechanism is that, under the latter interpretation, one must conclude
that firms essentially had no choice but to try to restore their market
power.
The problem with the purely behavioral interpretation is that it
provides no general explanation of how market structure is deter-
Price Competition and Concentration 147

mined for any given competition regime.47 Hence it cannot really be


seen as an alternative theory of the competition-concentration rela-
tionship. As a description of managerial behavior, however, it is
perfectly consistent with the structural mechanism outlined in this
chapter, and it may even help draw attention to an ancillary mecha-
nism that could potentially speed up the adjustment of concentration
to its new long-run equilibrium value following an increase in the
intensity of price competition. In particular, more efficient or finan-
cially stronger firms may try to restore their market power soon after
competition intensifies rather than go through a period of reduced
profitability while waiting for the exit of the more vulnerable firms—
which can take time, especially in industries with slow technological
progress and a slow rate of depreciation of the capital stock.
The mechanism that drives the evolution of structure in exogenous
sunk cost industries is relatively straightforward. Once we allow for
advertising or R&D, however, complications arise. The theoretical
predictions are no longer as clear and strong as the predictions for
exogenous sunk cost industries. More important, there has to be a
partial change of emphasis, from a single variable, concentration, to
the joint determination of a pair of variables, concentration and the
level of endogenous sunk costs (advertising or R&D). The next two
chapters examine the effects of competition in advertising-intensive
and R&D-intensive industries. Then the final chapter returns to some
of the most central predictions of the present theory, those that relate
to the joint determination of market structure and profitability, in a
way that highlights some of the basic similarities across classes of
industries.

47. Another problem with the behavioral interpretation is that it may not always be
valid. I will return to this issue in chapter 7, section 7.3.
This page intentionally left blank
5 Price Competition,
Advertising, and Market
Structure in Advertising-
Intensive Industries

5.1 Introduction

Chapter 4 provided strong econometric evidence to support the the-


oretical prediction of a positive effect of price competition on con-
centration in exogenous sunk cost industries. This chapter continues
the search for general theoretical results and empirical regularities on
the effects of competition, focusing this time on advertising-intensive
industries.
Things are more complicated in this case, because the intensifica-
tion of price competition may affect the incentive of firms to spend on
advertising. The first of two key issues examined in this chapter is
therefore the link between price and nonprice competition. The sec-
ond is the effect of competition on market structure. If more price
competition causes firms to spend less on advertising, then the effect
on market structure can, in principle, be ambiguous: the decrease in
the amount of sunk costs that each firm must cover at equilibrium
may or may not offset the fall in gross profits due to increased price
competition, so concentration may decrease or increase. Clearly, this
ambiguity arises only if more intense price competition is associated
with less advertising, a hypothesis that has received some empirical
support from case studies of particular industries (see Scherer and
Ross 1990, p. 673; Gasmi et al. 1992; Slade 1995), but on which there
is no systematic cross-industry evidence.
The objective of this chapter is, then, to analyze the joint effect
of short-run price competition on advertising and concentration in
advertising-intensive industries, combining theory, econometric anal-
ysis, and case-study evidence. The game-theoretic analysis brings
together and expands on two previous lines of research. The first is
150 Chapter 5

the literature on the determinants of market structure. Selten (1984)


and Sutton (1991) have examined how price competition affects
concentration in exogenous sunk cost industries, but they have not
discussed this issue in the context of industries with significant
endogenous sunk costs. The second is the literature on the interac-
tion between short-run and long-run decisions in oligopoly. Several
authors have examined the effect of anticipated short-run conduct on
strategic long-run decisions (e.g., Yarrow 1985, Ireland 1987, Jehiel
1992, Friedman and Thisse 1993, Fershtman and Gandal 1994, Ziss
1994, Fershtman and Pakes 2000). However, in most of these studies,
products are homogeneous or horizontally differentiated, and in
nearly all of them (the only exception being the study by Fershtman
and Pakes) market structure is taken as exogenous. In the model
described in this chapter, on the other hand, the product is vertically
differentiated (in the sense that there are several varieties of the
product, which differ in quality or ‘‘perceived quality’’) and the long-
run equilibrium is seen as involving the joint determination of mar-
ket structure and the level of advertising expenditure or intensity.
The exogenous factors in this framework are preferences, technology,
and short-run conduct.
In the next two sections I derive a number of theoretical predictions.
In advertising-intensive industries, theory places weaker constraints
on the space of outcomes than in exogenous sunk cost industries.
Nevertheless, it is possible to show that certain outcomes cannot
occur within the present framework, since they are inconsistent with
the requirement that net profits be zero in free-entry equilibrium,
while others, although possible, may be unlikely.
I then test the theory against empirical evidence on the effects of
the 1956 Restrictive Trade Practices Act on advertising and market
structure in advertising-intensive industries. I use econometric anal-
ysis to identify the overall effects of price competition across indus-
tries. This is, of course, a powerful test since it provides general
results and allows one to control for other factors that may have
influenced the evolution of industries during the period under study.
On the other hand, this approach may not fully reveal some of the
subtle ways in which price competition shapes the evolution of
advertising-intensive industries. To shed more light on the specific
mechanisms at work in these industries, the econometric analysis is
supplemented in this chapter by the analysis of case studies.
Price Competition, Advertising, and Market Structure 151

5.2 Theoretical Framework

Consider an advertising-intensive industry where each firm pro-


duces one variety of a differentiated product. Competition in such an
industry can be modeled by means of a three-stage game as follows.
At stage 1 each of N0 potential entrants decides whether or not to
enter the industry, and those firms that do enter pay an exogenously
given sunk cost of entry f . At stage 2 each firm i chooses to spend
an amount Ai on advertising. Advertising increases the consumers’
willingness to pay for the firm’s variety, either by increasing its
perceived quality or by providing information about the variety.1
Finally, at stage 3 firms set prices or quantities. Note that Ai is a fixed
cost which is incurred prior to the price or quantity competition
stage; in particular, a higher level of advertising is not associated
with any increase in the marginal cost of the firm. N0 is assumed to
be sufficiently large so that at any equilibrium of the three-stage
game there is at least one nonentering firm. In what follows, I will
characterize a symmetric subgame perfect equilibrium in pure strat-
egies in this three-stage game.
This game structure captures the idea that at any point in time
some variables are more difficult to change than others, and are
therefore taken as given when making shorter-term decisions. Hence
at any point in time some costs can be thought of as sunk, even
though firms in practice incur a continuous stream of expenditures.
A long-run strategic variable in every industry is the decision to enter
or not to enter. In industries where advertising is effective, and there-
fore nonprice competition through advertising is a key element in
the set of competitive strategies employed by firms, an additional
long-run strategic variable is the expenditure involved in creating an
advertising-based brand image. Note that whether or not advertising
should be regarded as a long-run decision has been a matter of some

1. In general, advertising will shift the demand function of a ‘‘representative’’ con-


sumer outward, so that the consumer is willing to pay a higher price for a given
quantity or to buy a larger quantity at the same price. This is what is meant by the
phrase ‘‘advertising increases the consumers’ willingness to pay.’’ Note that Ai could
also be interpreted as purely ‘‘informative’’ advertising if it were assumed to increase
market size, in the sense of an increase in the number of consumers, rather than the
willingness of any particular consumer to pay. The results of this section would still be
valid under such an interpretation, since the only crucial assumption in the present
framework is that Ai raises gross profit pi . The exact way this is brought about is
irrelevant.
152 Chapter 5

debate.2 Much recent work, however, has treated advertising as a


long-run decision on the grounds that brand image reacts relatively
slowly to current expenditure, is probably difficult to modify in the
short run, and is in all likelihood more difficult to modify than stan-
dard short-run choice variables, such as price or output. This chapter
follows this approach.
I assume, for simplicity, that advertising and the entry decision are
the only long-run choice variables in the game. It is, of course, true
that in some advertising-intensive industries R&D is also important.
R&D could also be included in the parameter Ai , in which case the
results of this section would be more generally valid for all classes of
industries with significant endogenous sunk costs (advertising or
R&D). Under this interpretation the parameter A would represent the
sum of advertising and R&D expenditure throughout this section.3
An alternative approach is to point out that the number of indus-
tries where advertising and R&D are both key strategic variables—as
suggested by an advertising-sales ratio and an R&D-sales ratio that
are both in excess of 2%, say—is in fact very small. Usually one or
the other of these two variables is dominant. Hence, in the present
chapter Ai can be treated as representing advertising only, on the
implicit assumption that the presence of endogenous R&D expendi-
tures in some advertising-intensive industries will not significantly
alter the mechanism and the results described. I adopt this approach,
although I will provide some further cautionary remarks on this issue
at the end of this section. Note that this approach is perhaps justified
by the fact that, for several of the key empirical results of this chapter,
the sample used consists of industries with typical advertising-sales
ratio higher than 2% (rather than 1%, which is the commonly used
cutoff point).

The Third-Stage Subgame

The equilibrium of the third-stage subgame can be represented by


a vector of payoffs pi ðS; N; h; e; t; A1 ; . . . ; Ai ; . . . ; AN ; g 1 ; . . . ; gi ; . . . ; gN Þ,

2. Schmalensee (1992) has argued that it is not clear that advertising should be seen as
a longer-term decision than price, because advertising depreciates relatively quickly
and price, too, can often have durable effects on demand. He also shows, however, that
key theoretical results derived from a three-stage game that treats advertising as a
sunk cost are robust to alternative approaches to modeling the interaction between
advertising and price competition, provided that advertising is treated as a fixed cost.
3. This is the approach taken in Symeonidis (2000b).
Price Competition, Advertising, and Market Structure 153

where S is market size, N is the number of firms that have entered at


stage 1, h is a measure of the degree of horizontal product differen-
tiation, e is a measure of ‘‘advertising effectiveness,’’ gi is a vector of
parameters specific to firm i and independent of N and the Ai ’s (such
as marginal cost or capacity), and t is a measure of the intensity of
price competition.
The interpretation of t is the same as in the previous chapter: for
any given N, pi will depend on the firms’ pricing strategies, which in
turn will partly depend on exogenous institutional factors, such as
the climate of competition policy or the degree of economic integra-
tion. Although a change in t will be mainly interpreted in this chap-
ter (as elsewhere in the book) specifically in terms of the degree of
collusion, it can be seen more broadly as any change in the gross
profit associated with firms’ conduct (but not with demand-side or
technological parameters such as h or e), for given N and Ai ’s. For
instance, a reduction in manufacturers’ margins, for given N and
Ai ’s, caused by pressure from retailers or competition from imports,
would represent an increase in the intensity of price competition t.
However, as pointed out in chapter 4, t is not equivalent to the price-
cost margin, which is treated here as endogenous.
In the benchmark case of symmetric single-plant firms, the equi-
librium gross profit of firm i can be written as pi ðS; N; h; e; t; Ai ; ðAi ÞÞ,
where ðAi Þ denotes the ðN  1Þ-tuple ðA1 ; A2 ; . . . ; Ai1 ; Aiþ1 ; . . . ; AN Þ.
Alternatively, one can write gross profit as pi ½S; CðN; Ai ; ðAi ÞÞ; h; e; t;
Ai ; ðAi Þ, where C is any concentration measure whose value
increases in 1=N for given Ai ’s (and depends only on N when Ai ¼ Aj ,
Ei; j; j 0 i), such as the n-firm concentration ratio. We assume that pi
is everywhere increasing in S, e, and h, and decreasing in t and N. In
addition, we make two assumptions about the effect on pi of changes
in own and rival advertising expenditure:

Assumption A1. pi is increasing and strictly concave in Ai with


dpi =dAi > 1 at Ai ¼ 0, and there exists a level of Ai such that
dpi =dAi ¼ 1 for any given N and set of Aj ’s.
Assumption A2. pi is nonincreasing in Aj , Ej 0 i.

For some of the results derived below, a third assumption is also


used:
Assumption A3. If Ai ¼ Aj ¼ A, Ei; j; j 0 i, and all the A’s increase
(decrease) by the same amount, pi cannot fall (rise).
154 Chapter 5

These assumptions are straightforward to interpret and seem


rather uncontroversial. The first one says that a firm’s gross profit is
increasing at a decreasing rate in own advertising expenditure, and
that there exists a point of the profit function, involving some positive
amount of advertising, such that one more extra unit of advertising
costs as much as the gross profit it creates.4 The second assumption
says that a firm’s gross profit is nonincreasing in any given rival’s
advertising expenditure. This does not rule out advertising spillovers,
but it does imply that any such spillovers are not very large.5 Finally,
the third assumption says that if, starting from a situation where all
firms spend the same amount on advertising, all firms equally in-
crease (decrease) their expenditure, then the gross profit of each firm
will not fall (rise). Essentially this implies that the effect of own ad-
vertising on own gross profit is sufficiently strong that it cannot be
offset by the joint effect of rivals’ advertising, when moving from one
symmetric configuration to another.
This last assumption is perhaps somewhat stronger than the first
two, so it may be worth pointing out that all three assumptions are
satisfied in standard models of vertical differentiation with symmet-
ric firms, such as the isoelastic demand model described in Sutton
(1991), and the linear demand model with quality indices described
in Sutton (1998) and also discussed later in this chapter. In any case,
assumption A3 is required only for propositions 5.2 and 5.3 below,
but not for proposition 5.1.

The Second-Stage Subgame

At stage 2 each firm chooses Ai to maximize its net profit pi  Ai  f ,


given the choices of the other firms and the number of firms that

4. Note that since dpi /dAi > 1 at Ai ¼ 0, firms always find it profitable to advertise.
One could also imagine that the function pi is first convex and then concave in Ai .
Although a firm would never optimally choose a point on the convex part of the gross
profit function representing a positive amount of advertising, it might choose not to
advertise at all, if the net profit from doing so were larger than the net profit at the
point where dpi =dAi ¼ 1. Since the focus of this chapter is on advertising-intensive
industries, it seems natural to ignore cases that involve zero advertising. I also abstract
from the issue of potential threshold effects (in the sense that advertising below a cer-
tain level has little impact), since any part of the gross profit function below the
threshold would have little relevance for the present analysis.
5. This point is further elaborated in the next chapter, section 6.2, where the issue of
spillovers is discussed in more detail in the context of R&D-intensive industries. The
key insights from that discussion are relevant for advertising spillovers as well.
Price Competition, Advertising, and Market Structure 155

have entered at stage 1. The first-order condition is dpi =dAi ¼ 1, Ei.


This says that each firm spends on advertising up to the point where
the cost of an extra unit of advertising is equal to the profit it creates
in the third-stage subgame. At any symmetric equilibrium, we have
Ai ¼ A, Ei, so the first-order condition for firm i can be written as
d
pi ðS; N; h; e; t; Ai ; ðAi ÞÞ ¼ 1 at Ai ¼ Ai ¼ A: ð5:1Þ
dAi
Equation (5.1) defines the level of advertising expenditure incurred
by firms as a function of the number of firms that have entered at
stage 1. Note that the assumption has been made that at stage 2 of
the game the Ai ’s are the Nash equilibrium outcomes of a non-
cooperative game, whatever the intensity of price competition at the
final stage. In other words, firms do not collude in advertising, either
perfectly or imperfectly. This assumption, which is common in the
theoretical literature on ‘‘semicollusion,’’ is largely motivated here
by the empirical evidence, according to which collusion with respect
to media advertising is generally rare (see Scherer and Ross, 1990,
chapter 16). The present empirical context is no exception to this:
collusion with respect to media advertising was certainly very rare in
British manufacturing industry in the 1950s, as discussed in chapter
2 of this book.

The Symmetric Equilibrium

At stage 1, firms enter as long as they can make a nonnegative net


profit, anticipating equation (5.1). The free-entry condition, which is
assumed to hold whatever the intensity of price competition at the
final stage of the game (in line with the empirical evidence on the
British cartels reviewed in chapter 2), is
pi ¼ pi ðS; N; h; e; t; Ai ; ðAi ÞÞjAi ¼Ai ¼A ¼ A þ f ; Ei; ð5:2Þ

where, for simplicity, N is treated as a continuous variable. This


implies that each firm makes exactly zero net profit at equilibrium.
The number of firms N  and level of advertising expenditure A  at
the symmetric equilibrium are defined by the two necessary condi-
tions (5.1) and (5.2). Since I am here concerned with comparative
statics results, I will assume that these two conditions define a
unique equilibrium with N  b 2 and A  > 0.
156 Chapter 5

There are a number of different effects at work following a change


in the intensity of price competition (or any other exogenous vari-
able). First, there is a direct effect on gross profit, given the initial
levels of A  and N  . Second, there is an effect on the firms’ incentive
to spend on advertising, given N  , which causes both gross profit
and the level of advertising expenditure (i.e., both sides of equation
(5.2)) to change. These changes also induce a change in N  until net
profit is driven back to zero. Finally, there are indirect effects of the
change in N  on profit as well as on advertising expenditure. Since
the sign and magnitude of these effects can vary, little can be said in
general about the comparative statics of A  and N  without impos-
ing more structure on the model. Clearly, there is no reason to expect
that more price competition will affect the incentive to advertise in
any particular way. Moreover, since both sides of the free-entry
condition will be affected by a change in t, it is not clear in what
direction N  should change to restore the free-entry condition.
But while the comparative statics properties of the model are am-
biguous when A  and N  are considered separately, several general
results can be derived for the joint behavior of A  and N  following a
change in the intensity of price competition.

The Joint Effect of Price Competition on Advertising and


Concentration

I begin with the first of three general results regarding the joint be-
havior of A  and N  (or the concentration ratio C  ) following a change
in the intensity of price competition t. Note that analogous results can
be derived for the effect of a change in any of the other exogenous
variables in the model. Let DA  and DN  ðDC  Þ denote the change
in A  and N  ðC  Þ, respectively. The following proposition places
bounds on the space of outcomes by excluding certain outcomes as
not admissible within the theory.

Proposition 5.1. Under assumptions A1 and A2, (1) an increase


in the intensity of price competition t cannot simultaneously cause
DA  b 0 and DN  b 0 ðDC  a 0Þ, and (2) no other outcomes are
excluded.
Proof. First consider part (1). The gross profit of firm i at the initial
equilibrium can be written as
Price Competition, Advertising, and Market Structure 157

p0 ¼ pðAi ¼ A0 j S; N0 ; h; e; t0 ; Ai ¼ A0 Þ ¼ A0 þ f ;

while the gross profit at the new equilibrium is


p1 ¼ pðAi ¼ A1 j S; N1 ; h; e; t1 ; Ai ¼ A1 Þ ¼ A1 þ f :

We will also need to define the profit of firm i at the initial equilib-
rium values of t and N  when firm i sets Ai ¼ A1 while all other firms
set Aj ¼ A0 :
p 0 ¼ pðAi ¼ A1 j S; N0 ; h; e; t0 ; Ai ¼ A0 Þ:
The proof is by contradiction. Assume that N1 b N0 and A1 b A0
following a rise in t (i.e., we also have t1 > t0 ). From A2 and the fact
that pi is decreasing in N and t, we obtain p1 < p 0 . Moreover, since
the function pi ðAi j S; N0 ; h; e; t0 ; Ai ¼ A0 Þ is concave in Ai and its
slope is equal to 1 at Ai ¼ A0 , its slope is smaller than 1 at all points
between A0 and A1 . Hence p 0  p0 < A1  A0 . Combining the two
inequalities, we obtain p1  p0 < A1  A0 . This is, however, impos-
sible, since net profit must be zero at equilibrium and therefore we
must have Dp  ¼ DA  . Hence it cannot be the case that DA  b 0 and
DN  b 0 following a rise in t.
To prove part (2) of the proposition, we only need to show that no
other outcomes can be excluded even for a much more structured
version of the model. The following set of ancillary assumptions is
intended to eliminate some of the indirect effects following a change
in t:
Assumption A4. dpi =dAi is nonincreasing in Aj , Ej 0 i.

Assumption A5. dpi =dAi is independent of N. This implies that A  is


independent of N at equilibrium.

Assumption A6. pi is homogeneous of degree zero in the Ai ’s. This


ensures that at equilibrium, when all firms set Ai ¼ A  , the payoff of
each firm in the third-stage subgame is independent of A  .

Consider now the effect of an increase in the intensity of competi-


tion on N  and A  . We distinguish three cases, depending on the sign
of the derivative dpi =dAi with respect to t. Suppose first that dpi =dAi
is everywhere increasing in t (i.e., that the return to a marginal in-
crease in Ai is higher when competition is more intense). In this case
an increase in t causes N  to fall and A  to rise. To see this, first no-
tice that when t increases, equation (5.1) no longer holds at the orig-
158 Chapter 5

inal level of A  . Assumptions A4 and A5 and the concavity of pi en-


sure that the new level of A  will be higher if (5.1) is to be satisfied.
The free-entry condition and assumption A6 then imply that, at the
initial level of N  , the equilibrium payoff of each firm will now be
lower than the advertising expenditure incurred. Hence N  must fall.
Next suppose that dpi =dAi is independent of t. It is then easy to see
that A  will not change and N  will fall following an increase in t.
Finally, if dpi =dAi is everywhere decreasing in t, the possibility arises
of more competition leading to a decrease in concentration. Equation
(5.1) no longer holds at the original level of A  once competition
intensifies. Assumptions A4 and A5 and the concavity of pi then en-
sure that A  falls. This may or may not offset the fall in pi caused by
the increase in t, so the direction in which N  must change to restore
the free-entry condition is not known. r

The intuition for proposition 5.1 is as follows. Starting from a zero-


profit equilibrium, an increase in t has a direct negative effect on net
profit. For the zero-profit condition to be restored, net profit must
rise. This cannot happen through a joint increase in the number of
firms (a fall in the concentration ratio) and advertising expenditure,
since this will unambiguously further reduce net profit. The reason is
that (1) entry reduces the profit of existing firms; (2) an increase in
own advertising has a stronger effect on sunk cost than on gross
profit when starting from a long-run equilibrium with dpi =dAi ¼ 1, Ei
(owing to the concavity of the gross profit function); and (3) an in-
crease in rival advertising cannot increase own profit.
Note that proposition 5.1 is a prediction on the joint effect of price
competition on firm advertising expenditure and market structure.
In empirical work, one often has to focus on industry advertising
expenditure or the advertising-sales ratio. What can the present
theory say about these variables? Consider first the industry adver-
tising expenditure N  A  . The key question is whether we can rule
out the case DðN  A  Þ b 0 and DN  b 0 ðDC  a 0Þ following an in-
crease in t. This joint outcome cannot result from DA  b 0 and
DN  b 0, because of proposition 5.1. And it is very unlikely to result
from DA  < 0 and DN  b 0, with the rise in N  offsetting the fall in
A  , so that N  A  rises. The reason is that the only way to obtain
DN  b 0 following a rise in t is through a significant fall in A  , as will
be explained in more detail below. But since a significant fall in A  is
required for even a small rise (or no change) in N  , it is very difficult
Price Competition, Advertising, and Market Structure 159

to obtain DðN  A  Þ b 0 together with DN  b 0 ðDC  a 0Þ following a


rise in t.
Next, consider the advertising-sales ratio. It seems plausible to as-
sume that a rise in the intensity of price competition t will not lead, in
general, to a fall in industry sales revenue in the long run. Hence,
ruling out the joint outcome DA  b 0 and DN  b 0, as proposition 5.1
does, also implies ruling out a joint increase in the advertising-sales
ratio (at the firm or at the industry level) and fall in concentration
under plausible assumptions about the effect of the rise in t on sales
revenue.
Proposition 5.1 suggests that only weak predictions can be made
regarding the effect of price competition on market structure and
advertising expenditure in advertising-intensive industries. This con-
trasts with the case of exogenous sunk cost industries, where the
competition effect on concentration is unambiguously positive, as
shown in chapter 4. The next proposition implies that while a nega-
tive effect of price competition on concentration in advertising-
intensive industries cannot be ruled out, it may be unlikely.
Proposition 5.2. Under assumptions A1–A3, a necessary condition
for N  to rise or remain unchanged (hence for C  to fall or remain
unchanged) following an increase in the intensity of price competi-
tion t is that equilibrium advertising expenditure A  falls, and this
fall is at least as large as the fall in gross profit directly caused by the
increase in t.
Proof. Suppose that equilibrium concentration falls or remains un-
changed following an increase in t. We know from proposition 5.1
that in this case A  must fall. We now want to show, in addition, that
the fall in A  must be at least as large as the fall in gross profit which
was the direct result of the increase in t. To be more precise, let p0
denote again the gross profit of firm i at the initial equilibrium and p1
the gross profit at the new equilibrium. The profit of firm i following
a rise in t at the initial levels of N  and A  is given by

p 00 ¼ pðAi ¼ A0 j S; N0 ; h; e; t1 ; Ai ¼ A0 Þ:


From proposition 5.1 we know that if a rise in t results in N1 b N0 ,
it must be the case that A1 < A0 . Also, the assumption that pi is
decreasing in t implies that p 00 < p0 . We now want to show that if a
rise in t results in N1 b N0 , then we must have A0  A1 b p0  p 00 .
160 Chapter 5

Assume the opposite, that is, A0  A1 < p0  p 00 , given that equi-
librium concentration falls or does not change following a rise in t.
From assumption A3 and the assumption that pi is decreasing in
N  , we obtain p1 a p 00 . Combining the two inequalities, we obtain
A0  A1 < p0  p1 , which is impossible, since net profit must be zero
at long-run equilibrium and hence we must have Dp  ¼ DA  . It fol-
lows that if concentration falls or does not change following a rise
in t, then it must be the case that A0  A1 b p0  p 00 . This completes
the proof. Note that the reverse is not true, that is, the condition
A0  A1 b p0  p 00 is not sufficient to rule out an increase in concen-
tration following a rise in t. (For instance, concentration may rise if
the fall in A  reduces gross profit below p 00 .) r

Proposition 5.2 is driven by the fact that a ‘‘small’’ fall in advertis-


ing following a rise in t would not be sufficient to offset the direct
negative effect of the rise in t on gross profit, and hence net profit
would fall at the initial level of concentration. Thus concentration
would have to rise to restore the free-entry condition through a rise
in profit. To obtain a fall (or no change) in concentration, the fall in
A must be large enough to offset the effect of the rise in t on gross
profit, so that net profit rises (or does not change) at the initial level
of concentration.
Proposition 5.2 is a conditional prediction on the competition-
concentration relationship. Since the prediction is conditional on the
behavior of another observable variable, profit, it is empirically test-
able. In particular, an empirical test of this proposition can take two
forms. If concentration does not rise (the number of firms does not
fall) in the long run following an intensification of price competition
in an advertising-intensive industry, then firm advertising expen-
diture must have fallen and the fall must be at least as large as the
fall in gross profit directly caused by the intensification of competi-
tion. Alternatively, if firm advertising expenditure rises, remains
unchanged, or falls by less than the fall in gross profit directly caused
by the intensification of price competition, then concentration must
rise (the number of firms must fall) in the long run.
Like proposition 5.1, proposition 5.2 relates to advertising expen-
diture at the firm level, but also has indirect implications for industry
advertising expenditure. Broadly speaking, the proposition says that
if concentration does not rise when price competition intensifies,
then it must be the case that firm advertising has fallen a lot. It is
Price Competition, Advertising, and Market Structure 161

then also likely that industry advertising and the advertising-sales


ratio will have fallen a lot. On the other hand, if this latter effect is
not observed when price competition intensifies, then concentration
must rise.
Although proposition 5.2 involves a comparison between endoge-
nous variables, these are related to exogenous industry characteristics.
Define r as the fraction of profit that is lost as a direct consequence of
the rise in t, that is, ð1  rÞp0 ¼ p 00 , where p0 has previously been
defined as the gross profit at the initial equilibrium and p 00 has been
defined as the gross profit following the rise in t at the initial levels of
A  and N  . Recall that f > 0 denotes the exogenous element of sunk
cost.

Proposition 5.3. Under assumptions A1–A3, a necessary condition


for N  to rise or remain unchanged (hence for C  to fall or remain
unchanged) following an increase in the intensity of price competi-
tion t is that the fraction of endogenous sunk costs in total sunk costs
at the initial equilibrium is larger than the fraction of gross profit lost
as a direct result of the rise in t, namely, that 1  f =p0 > r.
Proof. According to proposition 5.2, if concentration does not rise
when t increases, then A0  A1 b p0  p 00 . Because of the zero-profit
condition p0 ¼ f þ A0 , the previous inequality implies p 00 b f þ A1 ,
and therefore p 00 b f , which is equivalent to 1  f =p0 > r. r
In industries where exogenous sunk costs are an important part of
total sunk costs, the free-entry condition implies that the profit margin
can be significantly larger than advertising intensity. Hence, when
price competition intensifies, there may be little scope for a fall in A 
that matches the fall in gross profit. Proposition 5.3 implies that a
nonpositive effect of price competition on concentration may be un-
likely in industries with substantial exogenous sunk costs and rela-
tively low advertising intensity, unless the intensification of price
competition has only a small direct effect on the profit margin.6

6. A numerical example may help clarify this point. Consider an industry with
symmetric firms and unit-elastic demand, so that the industry sales revenue is fixed
and equal to S. Let ð p  cÞ=p denote the price-cost margin and NA/S the industry
advertising-sales ratio. The free-entry condition implies ð p  cÞ=p ¼ NA=S þ Nf =S,
where f is the exogenous sunk cost.
Suppose that initially ð p  cÞ=p ¼ 6% and NA=S ¼ 1%. Now if the price-cost margin
falls below 5% as a direct result of tougher price competition, concentration must rise
because A cannot fall below zero. In other words, at any new equilibrium we must
162 Chapter 5

Remarks and Clarifications

Several remarks on the general results derived above are in order.


First, the parameter A has been interpreted throughout this section
as expenditure on advertising. Thus the implicit assumption has been
made that R&D is not very important in most advertising-intensive
industries. Now the number of industries where advertising and R&D
are both key strategic variables is indeed small; nevertheless, such
cases exist. The question then arises as to how the presence of sig-
nificant endogenous R&D expenditures in these industries may affect
the results presented here. The answer is that propositions 5.1–5.3
would still hold, but the parameter A would have to be interpreted
as the total amount of endogenous sunk costs, that is, the sum of
advertising and R&D costs. Provided one adopts this interpretation,
propositions 5.1–5.3 would be valid as general statements on the
links between price competition, endogenous sunk costs, and market
structure in industries where one or more types of endogenous sunk
costs are significant.
Second, the results of this section have been derived for the bench-
mark case of symmetric single-product firms. Extending the anal-
ysis to the case of asymmetric or multiproduct firms is beyond my
present scope. However, it is not clear what other effects may arise
in these more complex settings that would offset the mechanisms
identified for the benchmark case. Thus it is certainly the case
that advertising-intensive industries typically comprise multiproduct
firms with very different advertising intensities, including firms that
advertise very little or not at all. However, low advertisers usually
have low sales, so they do not significantly affect either the concen-
tration ratio or the industry advertising-sales ratio (although they
may affect the total number of firms). Moreover, certain choices in-

have NA=S b 0, so the free-entry condition can be restored only through a rise in the
price-cost margin and/or a fall in Nf/S, both of which require a fall in N. Note that
even with a smaller fall in the price-cost margin, a large proportional change in the
advertising-sales ratio is required if concentration is not to rise.
Things are different for high-advertising industries. Thus, for instance, if initially
ð p  cÞ= p ¼ 6% and NA=S ¼ 4%, then a fall of the price-cost margin to 5% can be easily
offset by a fall in A, and concentration need not rise. In particular, if the price-cost
margin is homogeneous of degree zero in the Ai ’s, then N will remain unchanged pro-
vided that NA/S goes to 3%, which represents only a 25% decrease in the advertising-
sales ratio; and it will rise if A falls by more than 25%. If, on the other hand, a fall in A
causes a reduction in the price-cost margin, then A will need to fall by more than 25%
to keep N constant; but this will still be a modest proportional fall in A.
Price Competition, Advertising, and Market Structure 163

volving multiple varieties, such as the introduction or promotion of


secondary brands by high advertisers, may be treated for some pur-
poses as analogous to advertising choices of single-product firms.
Finally, the presence of efficiency differences should not matter more
in advertising-intensive industries than in exogenous sunk cost in-
dustries, and it has already been argued in chapter 4 that the basic
theoretical insights regarding the competition effect on market struc-
ture are not substantially modified in the presence of efficiency
differences.
In any case, the question of whether the theoretical results derived
in this section are indeed those that matter for analyzing the effects
of price competition in advertising-intensive industries is ultimately
an empirical one. The theory developed will be tested below against
econometric evidence as well as against evidence from case studies
of particular industries.
Third, the results of this section were derived from a model where
advertising increases the consumer’s willingness to pay rather than
providing information about the product. In the empirical analysis,
however, the data used are for total advertising, and no distinction is
made between ‘‘persuasive’’ and ‘‘informative’’ advertising. How-
ever, this discrepancy is more apparent than real. First, as mentioned
above, the basic theoretical results of this section would be valid
even if advertising were purely informative. Second, in most brand
advertising by manufacturers the ‘‘persuasive’’ element is dominant,
especially in consumer-good industries, which represent the bulk of
the samples used in this chapter.
Finally, let me point out once again that the assumption of the
exogeneity of the intensity of price competition t is a simplification.
The arguments developed in chapter 4 are valid here as well. To
summarize, the assumption of exogeneity of the theoretical variable t
is probably justified in the present case because, first, cartelization in
the 1950s was largely a function of exogenous industry character-
istics and, second, the key determinant of changes in firms’ pricing
conduct during the period under study was the exogenous change in
cartel policy. Moreover, even if t is seen as endogenous (i.e., as a
function of an exogenous institutional variable T and a vector of
other variables, possibly including N and A), the results of this sec-
tion could be restated as predictions on the behavior of N  and A 
conditional on the observed change in t, following a change in the insti-
tutional variable T. One can therefore conduct a reduced-form anal-
164 Chapter 5

ysis of the effect on market structure and nonprice competition of a


change in the main determinant of t during the period under study,
namely, the institutional variable T.

5.3 A Specific Example: The Linear Demand Model with Quality

To gain additional insight on the links between price competition,


advertising, and market structure in advertising-intensive industries,
I now briefly examine a model of vertical product differentiation
with endogenous market structure that can be seen as a specific case
along the lines of the theoretical framework developed in the previ-
ous section. This is in fact an extended version of the linear demand
model introduced in chapter 4. It should be emphasized from the
outset that some of the results of this model are probably specific to
its particular structure and the functional forms used. The interest of
the model for my present purposes lies in the fact that, despite its
specificity, it is sufficiently flexible to generate several of the possible
outcomes of the more general theoretical framework of the previous
section. As a result, the model can provide additional insight as to
the kinds of factors that determine these outcomes.

The Basic Model

Consider an advertising-intensive industry producing a potentially


infinite number of varieties of a differentiated product. Competition
in the industry is described by a three-stage game as follows. There
are N0 potential entrants, each with the capacity to produce a single
variety of the product. At stage 1 they decide whether or not to enter
at an exogenously given sunk cost of entry f . At stage 2 each firm
chooses a variety, described by a vertical attribute u that I will call
‘‘quality’’ and, in the case of an advertising-intensive industry, rep-
resents brand image. Quality increases the consumers’ willingness to
pay for the firm’s variety, but it comes at a sunk cost—namely, ad-
vertising expenditure—which is incurred at stage 2. Finally, at stage
3 firms set quantities. Assume that N0 is sufficiently large so that at
any equilibrium of the game there is at least one nonentering firm.
There are S identical consumers, and the utility function of each
consumer takes the form
!
X xi2 X X x i xj
U¼ xi  2  s þ M: ð5:3Þ
i
ui u u
i j<i i j
Price Competition, Advertising, and Market Structure 165

This is the quadratic utility function used in chapter 4, augmented


by the quality indices ui (see Sutton 1996, 1997b, 1998; Symeonidis
2000c). The parameters a1 and a2 have been dropped to ease nota-
tion; these are scale parameters and do not affect the comparative
statics results derived from the model. Recall that s, s A ð0; 2Þ, is an
inverse measure of the (exogenous) degree of horizontal product
differentiation. We require that ui b 1, Ei; this implies that advertis-
ing expenditure takes nonnegative values (see below).
The consumer’s inverse demand for variety i is given by
2x i s X xj
pi ¼ 1   ð5:4Þ
ui2 ui j0i uj

in the region of quantity spaces where prices are positive, and her
demand function for variety i is
P
ui ð1  pi Þ½2 þ sðN  2Þ  s uj ð1  pj Þ
xi j0i
¼ ð5:5Þ
ui ð2  sÞ½2 þ sðN  1Þ
in the region of prices where quantities are positive, where N denotes
the number of firms that have entered the industry. It can be easily
seen that x i is decreasing in pi and increasing in pj , Ej 0 i. Also, it is
increasing in ui and decreasing in uj , Ej 0 i.7 Since there are S con-
sumers, firm i sells quantity Sx i . Let each firm have a constant mar-
ginal cost of production c, where c < 1.
As in chapter 4, I will use here a continuous reduced-form collu-
sion parameter to model different degrees of competition in the final
stage of the game. In particular, I assume that in the final stage each
firm maximizes the sum of its own (gross) profit and a fraction l of
the (gross) profit of each of the rival firms. The parameter l, 0 a
l a 1, is, then, an inverse measure of the intensity of short-run com-
petition, with l ¼ 0 representing the Cournot-Nash equilibrium and
l ¼ 1 corresponding to perfect collusion.
Although the use of l is a somewhat arbitrary way to model col-
lusion, it turns out to be very useful in providing insight about the
effects of price competition in advertising-intensive industries within

7. A nice feature of the present model—and one that distinguishes it from models of
‘‘pure vertical differentiation,’’ which do not have this property—is that it gives rise to
reduced-form profit functions where profit increases in own quality and decreases in
rival quality.
166 Chapter 5

the context of the present example.8 Note that, as in chapter 4, the


assumption of exogeneity of l is justified here not only by an appeal
to well-known theoretical results of a multiplicity of equilibria in
models of repeated games, but also by reference to the empirical
context of this book: when the principal influence on l is a change in
the institutional framework, it is plausible to assume that any feed-
back effects from N or the ui ’s on l are small (i.e., ql=qN A 0 and
ql=qui A 0, Ei) and can be ignored.
Finally, we need to specify an advertising cost function linking
quality ui to advertising expenditure Ai . Let this function be given by
Ai ¼ eðuib  1Þ, where e is a positive constant and b is a measure of the
cost of advertising or, alternatively, an inverse measure of the effec-
tiveness of advertising in raising a product’s perceived quality.9 We
require that b > 2; this amounts to assuming that AðuÞ rises with u
more rapidly than gross profit and ensures that the model is well
behaved in a sense that will become clear below. (For instance, b > 2
is a necessary condition for the existence of a symmetric equilibrium
in the three-stage game with at least two firms in the market.)10 Since
ui b 1, it follows that Ai b 0. Note that firms are assumed to choose
ui (and hence Ai ) at stage 2 of the game to maximize own profit, ir-
respective of the value of l.
Firms must also incur an entry cost f at stage 1 of the game. This
may include some minimal advertising expenditure that is necessary
for producing a product of the lowest quality ðu ¼ 1Þ. To simplify the
analysis, it is assumed in what follows that f ¼ e, so the total sunk
cost incurred by firm i prior to the final stage is Fi ¼ Ai þ f ¼ euib .
This simplification is particularly justified for industries with high

8. The properties of l in the final-stage subgame are the same as in the version of the
model without quality indices described in chapter 4: for given N and for ui ¼ u, Ei, the
equilibrium price, price-cost margin, and profit in the second-stage subgame increase
and the equilibrium quantity falls as l rises. None of the other exogenous variables
that affect gross profit, namely, s or c, have properties similar to those of l.
9. No attempt is made here to explain what determines b and why it differs across
industries. Advertising effectiveness is simply taken as an industry-specific character-
istic which is not affected by firms’ strategies. This is consistent with the fact that the
industries where advertising ‘‘works’’ tend to be fairly similar across countries and
over time.
10. As pointed out in Sutton (1998), the economics of this model depends only on the
composite mapping from firms’ fixed costs to firms’ gross profits, rather than on the
separate mappings of fixed costs to qualities and from qualities to gross profits. Thus
the results would not change if we replaced u with u g , g > 0, in the utility function and
required that b > 2g.
Price Competition, Advertising, and Market Structure 167

advertising intensity, that is, when f represents a small fraction of


the total sunk cost incurred.11

The Symmetric Equilibrium

A symmetric subgame perfect equilibrium in pure strategies in the


three-stage game can be computed following the methodology of
Sutton (1991). The two necessary conditions for a symmetric equi-
librium in which N firms offer goods of quality v > 1 are
 
dpi  dF 
¼ ; Ei; ð5:6Þ
dui ui ¼v du u¼v

and

pðvjN; vÞ ¼ FðvÞ; ð5:7Þ


where N is assumed to be a continuous variable. Equation (5.6)
determines the equilibrium level of u conditional on N (i.e., the equi-
librium of the subgame starting at the second stage of the game),
while (5.7) is a free-entry condition. It can be shown that, provided
that S is sufficiently large (or e is sufficiently small) and b is sufficiently
large, there exists a unique symmetric subgame perfect equilibrium
of the three-stage game with N  b 2 and v  > 1 (see Symeonidis
2000c).12 This is defined by equations (5.6) and (5.7) after substituting
the values of p and F.
In particular, from (5.6) we obtain

Sð1  cÞ 2 ½E3 ðE1  E2 Þ þ E1 ðE3  E4 Þ


ðv S Þ b2 ¼ ; ð5:8Þ
be½4  sð1 þ lÞ 2 ½4 þ sð1 þ lÞðN  1Þ 2

which defines the equilibrium level of quality conditional on the


number of firms, v S ðNÞ, where

11. It is easy to check that for any given f 0 e, Ai þ f ! euib as ui increases.


12. When S is not sufficiently large (for given e), then a symmetric equilibrium in the
three-stage game with v  > 1 does not exist. In this case firms enter with the lowest
possible quality level u ¼ 1 and a symmetric equilibrium is derived from a two-stage
game with entry at a sunk cost f at stage 1 and quantity competition at stage 2 (and no
endogenous sunk costs). This is the game analyzed in chapter 4. The intuition for this
case is that firms do not advertise when market size is small because the gross profit
generated is not enough to cover the fixed advertising cost. Note that the model of
chapter 4 can also be derived as a special case of the present model when b is very
large, that is, advertising effectiveness is very small (and S and e are kept fixed).
168 Chapter 5

E1 ¼ 2½4 þ sð1 þ lÞðN  2Þ  ls 2 ð1 þ lÞðN  1Þ;

E2 ¼ 2sð1  lÞðN  1Þ;


E3 ¼ 4 þ sð1 þ lÞðN  2Þ;

E4 ¼ sð1 þ lÞðN  1Þ
are all nonnegative expressions. From (5.7), and using also (5.8), we
obtain
E3 ðE1  E2 Þ þ E1 ðE3  E4 Þ ¼ b½2 þ lsðN  1Þ½4  sð1 þ lÞ 2 ; ð5:9Þ
which defines the equilibrium number of firms N  . Given N  , (5.8)
then gives the equilibrium level of quality in the three-stage game,
v  ¼ v S ðN  Þ. Note that N  is independent of S, c, and e. This is due to
the simplifying assumption f ¼ e. With a more general function FðuÞ,
N  could be positively or negatively related to S, c, and e. This as-
sumption allows us, however, to focus on the variables of interest,
that is, l, b, and s.

Comparative Statics

The key comparative statics results of this model, within the range
of parameter values for which a symmetric equilibrium with N  b 2
and v  > 1 exists, have been analyzed elsewhere (Symeonidis 2000c).
They can be briefly summarized as follows.
First, a fall in l (more intense short-run competition) unambig-
uously reduces quality, and hence advertising expenditure. This is
true whether N is kept fixed—as can be easily checked by differen-
tiating v S ðNÞ with respect to l—or is allowed to adjust to its equilib-
rium value according to equation (5.9). Thus more price competition
implies less nonprice competition. While this result may be spe-
cific to the linear demand model, it is precisely because of this result
that the possibility of an ambiguous effect of a change in l on N 
arises.
Indeed, the second key result is that a fall in l may cause either an
increase or a decrease in N  , depending on the values of b and s and
the initial value of l itself. The ambiguity of the competition effect on
concentration in this model stems from the fact that more intense
price competition reduces advertising expenditure: since a fall in l
causes both gross profit and sunk cost to decrease, for given N, it is
Price Competition, Advertising, and Market Structure 169

not clear in which direction N  must change to restore the zero-profit


condition.
It is also worth mentioning a third result of the model: a decrease
in the degree of horizontal product differentiation (a rise in s) raises
advertising expenditure and reduces the number of firms in long-run
equilibrium. In other words, l and s affect the endogenous variables
of interest in very different ways. Thus a rise in s and a fall in l both
reduce price and gross profit, for any given N. But the former raises
advertising expenditure and lowers the number of firms in the long
run, while the latter lowers advertising expenditure and has an am-
biguous effect on the number of firms. This is worth emphasizing in
view of the fact that s (or a similar index of product substitutability)
is often used as a measure of competitive pressure in theoretical
studies, on the grounds that a rise in s reduces profitability for any
given number of firms. The present model suggests that it is impor-
tant to distinguish between the various sources of competitive pres-
sure when analyzing industries with significant endogenous sunk
costs.
For my present purposes, also of particular interest are the com-
parative statics properties of the model with respect to l in the
neighborhood of the Cournot-Nash equilibrium (i.e., for values of l
close to 0). As it turns out, the effect of a change in l on N  can then
be either positive or negative. The advantage of focusing on this polar
case is that we can compare the two possible outcomes of the model
for the same range of values of l, and thus highlight the role of
parameters other than l, such as s and b, in determining the sign of
the competition effect on market structure.
Proposition 5.4. In the linear demand model with quality indices,
an increase in the intensity of competition (a fall in l) decreases
(increases) the equilibrium number of firms N  in the neighborhood
of the Cournot-Nash equilibrium if b is greater (smaller) than 2 þ
½8s=ð4  sÞ 2 .

Proof. From the total differential of (5.9) we can compute dN  =dl as


a function of N  . Evaluate the resulting expression at l ¼ 0. Then
substitute into this expression the equilibrium value of N  for l ¼ 0.
We obtain

dN   ðb  2Þ½8s þ ðb  2Þð4  sÞ 2 
¼ ;
dl l¼0 16s
170 Chapter 5

which, given b > 2, is positive for b > 2 þ ½8s=ð4  sÞ 2 . By continuity,


the condition dN  =dl > 0 , b > 2 þ ½8s=ð4  sÞ 2  also holds in the
neighborhood of the Cournot-Nash equilibrium. r

Note that the function gðsÞ ¼ 2 þ ½8s=ð4  sÞ 2  is an increasing


function of s in the interval ð0; 2Þ. For s ! 0, gðsÞ ! 2, while for
s ! 2, gðsÞ ! 6. Thus, for values of b b 6 the competition effect on
concentration is always positive in the neighborhood of the Cournot-
Nash equilibrium. For b < 6 the competition effect on concentration
can be positive or negative, depending on the degree of horizontal
product differentiation: for any particular value of s, say s0 , there is a
corresponding value of b, say b0 (where b 0 < 6), such that the com-
petition effect on concentration is negative for all b A ð2; b0 Þ and posi-
tive for all b A ðb0 ; 6Þ. Now recall that the higher the value of b, the
lower the effectiveness of advertising in any particular industry.
What the above results therefore imply is that, in the neighborhood
of the Cournot-Nash equilibrium, a negative competition effect on
concentration is possible only in industries with high advertising ef-
fectiveness. These are also, other things being equal, industries with
high advertising intensity. Moreover, it is precisely in these indus-
tries that advertising expenditure falls a lot when price competition
intensifies in the neighborhood of the Cournot-Nash equilibrium, as
the next result shows.
Proposition 5.5. In the linear demand model with quality indices,
an increase in the intensity of competition (a fall in l) in the neigh-
borhood of the Cournot-Nash equilibrium has a stronger effect on
firm advertising expenditure A  ¼ eðvb  1Þ, the higher the effec-
tiveness of advertising (the lower the value of b).
Proof. Differentiate A  ¼ eðvb  1Þ with respect to l and evaluate
the resulting expression at l ¼ 0. Then substitute into this expression
the equilibrium value of N  for l ¼ 0. This yields
 " #
2 b=ð b2Þ
dA   ð bþ4Þ=ð b2Þ 2 Sð1  cÞ
¼2 bðb  2Þð4  sÞ e
dl l¼0 e

 ½2s þ bð4  sÞð3b2Þ=ð b2Þ :


Note that this expression is positive for all values of b and s in the
relevant ranges. Now differentiate this with respect to b to obtain
Price Competition, Advertising, and Market Structure 171

"  #
q dA  
qb dl l¼0

2ð bþ4Þ=ð b2Þ ð4  sÞ 2 eE5


b=ð b2Þ
½E7 þ 6bE6 ln 2 þ 2bE6 lnðE5 =E62 Þ
¼ 4ð b1Þ=ð b2Þ
;
ðb  2ÞE6

where
E5 ¼ Sð1  cÞ 2 =e;

E6 ¼ 2s þ bð4  sÞ,
E7 ¼ ðb  2Þ½ðb 2 þ 2sÞð2  sÞ þ 2ðb  sÞ 2 

are all positive expressions for b > 2, s A ð0; 2Þ. Moreover, it can be
checked from equations (5.8) and (5.9) that ðv  Þ b2 jl¼0 ¼ E5 =E62 . Since
we require v  > 1, we also have lnðE5 =E62 Þ > 0. Hence ðdA  =dlÞjl¼0 is
decreasing in b. By continuity, dA  =dl is decreasing in b in the
neighborhood of the Cournot-Nash equilibrium. r

The mechanism driving these results is similar to the mechanism


driving propositions 5.2 and 5.3. In industries with low advertising
intensity there is little scope for a fall in advertising expenditure that
matches the fall in gross profit, when price competition intensifies.
So a negative (or nonpositive) effect of price competition on concen-
tration is unlikely in these industries: the fall in advertising following
an intensification of short-run competition will probably be too small
relative to the fall in gross profit, so that net profit will decrease at the
initial level of concentration, and concentration will have to rise to
restore the free-entry condition. On the other hand, a negative (non-
positive) competition effect on concentration can occur if advertising
expenditure falls a lot relative to gross profit, so that net profit rises
(does not fall) at the initial level of concentration. This can happen
more easily in an industry where advertising intensity is initially
high.

Summary of the Theory and Implications for the Empirical Analysis

To summarize, the linear demand model with quality indices illus-


trates quite nicely, and develops further, some of the main theoretical
results on the effect of price competition on advertising and market
structure in advertising-intensive industries. The key theoretical pre-
172 Chapter 5

dictions of this chapter are, then, as follows. First, an increase in the


intensity of competition cannot simultaneously cause concentration
to fall (or not to change) and advertising expenditure to rise (or not
to change). Second, no clear results can be derived on the effect of
price competition on advertising without imposing a considerable
amount of structure on our theoretical models. In one such example,
the linear demand model, we have obtained the unambiguous result
that more price competition reduces nonprice (advertising) competi-
tion. Third, while a negative effect of price competition on concen-
tration cannot be ruled out, it is unlikely, since it can occur only if
there is a significant fall in advertising expenditure. If, on the other
hand, advertising falls little or does not change or rises following an
increase in the intensity of price competition, then concentration
must rise.
The theory should now be confronted with the empirical evidence.
Several questions can be asked. First, is the evidence consistent with
proposition 5.1, that is, is it the case that we do not observe the out-
comes that are excluded under this proposition? Second, what was
the overall effect of price competition on advertising across industries
in the present case? Third, is there any evidence of an unambiguous
overall effect of price competition on concentration in advertising-
intensive industries despite the inconclusiveness of the theory? Fourth,
is the experience of particular industries consistent with the condi-
tional predictions on concentration (propositions 5.2 and 5.3)?

5.4 The Data

The econometric analysis of the effects of price competition in


advertising-intensive industries in this chapter makes use of several
different samples of industries in separate regressions for concentra-
tion and advertising intensity. The reason for using different samples
for each of the endogenous variables is that the industry definitions
across the various statistical sources are often difficult to match.
Many of the variables and the data sources are the same as those
used in chapter 4 and will not be discussed again here. This is the
case for the concentration ratio, the setup cost proxies (capital stock
divided by the number of plants or the capital-labor ratio), the market
size proxies (deflated sales revenue), and the competition variable
CHANGE. There is a small difference in the definition of sales revenue
between concentration and advertising regressions: in the former case
Price Competition, Advertising, and Market Structure 173

I use sales revenue by UK firms, while in the latter case I use sales
revenue in the UK market.
The new variables in this chapter are advertising intensity and a
set of variables that capture differences across industries and over
time in the fraction of TV advertising in total advertising. To con-
struct these variables, systematic data on advertising expenditure
across a range of advertising-intensive industries were required.
These data are described in some detail below.
Advertising-sales ratios were constructed for 1954, 1958, 1963,
1968, and 1973. The first four of these years are the only ones during
the 1950s and the 1960s for which comparable sales revenue data
across industries are available at a fairly disaggregated level. On the
other hand, 1973 is the last year before the recession of the mid-
1970s, which seems to have caused a temporary but significant fall in
advertising expenditure in many industries. The data used in the
concentration regressions are, as in chapter 4, for 1958, 1963, 1968,
and 1975. Recall that 1954 and 1958 are ‘‘before’’ dates in the natural
experiment, 1963 is a year when we expect to see only the short-run
effects of competition, 1968 is a year when much of the long-run ef-
fect should have been realized, and by 1973 or 1975 the full long-run
effects of competition had certainly been realized.

Defining the Samples of Advertising-Intensive Industries

This chapter focuses on advertising-intensive industries, defined as


those industries with an average or typical advertising-sales ratio
(ADS) of more than 1% over the relevant period. The 1% cutoff point
was chosen because it is commonly used to classify industries ac-
cording to their advertising intensity. In addition, I have defined two
subsamples of high-advertising industries, one for the concentration
regressions and one for the advertising regressions. Each of these
includes industries with average or typical ADS higher than 2% over
the relevant period. These subsamples were defined in order to test
some of the theoretical results using a group of industries for which
advertising is clearly a key strategic variable. This may not always be
the case for industries with ADS between 1% and 2%; in fact, indus-
tries with medium advertising intensity may have features of exoge-
nous sunk cost industries.
The choice of the 2% cutoff point was the result of a trade-off. On
the one hand, the subsamples of high-advertising industries should
174 Chapter 5

contain a nonnegligible number of industries with a change in com-


petition regime, and hence the cutoff point should not be set too
high. On the other hand, they should be sufficiently different from
the ADS > 1% samples, and so the cutoff point should not be set too
low. As it turned out, the 2% cutoff point satisfies both these criteria.
In addition, it splits the group of advertising-intensive industries
used in the concentration regressions into two groups of equal size:
of fifty-six industries with ADS > 1%, twenty-eight had ADS > 2%.
For the samples used in the concentration regressions, the classifi-
cation of industries with respect to these two cutoff points was done
in the manner described in chapter 4. Thus I estimated the average or
typical ADS of all four-digit industries for which concentration data
were available, and then checked whether these were higher or lower
than 1% and 2%. For this purpose, advertising expenditure was
defined as media advertising, that is, other selling expenses were not
included on the grounds that they are generally incurred at the price
competition stage and do not create any ‘‘goodwill.’’13
For the samples used in the advertising regressions, the classifica-
tion of industries with respect to their typical ADS using the 1% and
the 2% cutoff points was done in a slightly different way. One com-
plication in this case was that the 1954 figures were usually untypi-
cal, because many British industries did not fully emerge from the
implications of wartime controls until the mid-1950s, and also be-
cause TV advertising was not introduced in the UK until 1955. On
the other hand, the 1956 legislation did not have any impact prior to
1958 and, in fact, its impact was largely realized between 1958 and
1968. I therefore used only the 1958, 1963, and 1968 advertising-sales
ratios to classify the industries according to their typical advertising
intensity. I did this by simply taking the average ADS for the three
‘‘core’’ years and comparing it with the two cutoff points.

Constructing Advertising-Sales Ratios

The advertising-sales ratio is used in this chapter not only to classify


industries as advertising-intensive (ADS > 1%) or high-advertising
(ADS > 2%), but also as an endogenous variable. Constructing this

13. Other selling expenses include ‘‘direct selling activities’’ such as visits by sales-
persons, which are common in producer-good industries, and ‘‘promotional activities’’
such as coupons to consumers and special terms to retailers, which are common in
consumer-good industries.
Price Competition, Advertising, and Market Structure 175

variable involved matching three different industrial classifications.


Data on advertising expenditure in the UK market, separately for
press and TV, were obtained from the Statistical Review of Press Ad-
vertising and the Statistical Review of Independent TV Advertising,
both published by Legion Information Services, Ltd., until 1963; the
Statistical Review of Press and TV Advertising, published by Legion
Information Services, Ltd., from 1963 until the early 1970s; and the
MEAL Monthly Digest of Advertising Expenditure, published since 1968.
These data are reported both for individual brands and for indus-
tries, at a relatively low level of aggregation, but are often not avail-
able for low-advertising industries. Data on sales revenue of UK
manufacturers at current net producer prices were obtained from the
individual industry reports of the Census of Production (various
years) and the 1973 issues of Business Monitor, PQ series: Quarterly
Statistics. These figures relate to UK firms rather than the UK market,
however. Sales revenue figures for the UK market were derived by
adding to the Census sales revenue figures the value of retained
imports in the UK and subtracting the value of exports by UK firms. I
also subtracted from the sales figures any amount of customs and
excise duties. Import and export data were taken from the Annual
Statement of the Trade of the United Kingdom.14
The choice of industry definitions for the advertising regressions
was dictated by theoretical as well as practical considerations. The
former included using an appropriate level of aggregation from the
point of view of the possible existence of advertising spillovers across
different brands of a single firm and also from the point of view of
competition. The latter included matching different industrial classi-
fications and ensuring comparability in the light of changes over time
in some of the industry definitions used in various data sources. In
most cases my industry definitions were based on the categories
used in the advertising statistics, which roughly correspond to the

14. A minor problem was that while the Census sales revenue data are for plants (for
1968 and 1973) or firms (for 1954, 1958, and 1963) employing at least twenty-five per-
sons, the import and export statistics cover all firms. Since very small firms often pro-
duce ancillary rather than core industry products and may spend little on advertising
or advertise only locally, the exclusion of very small firms (or plants) is a minor issue
for many industries, especially when imports and exports are small. In cases where
imports or exports were significant, or very small UK producers accounted for a con-
siderable fraction of total sales, an approximate adjustment of the sales figures was
made using published estimates of sales by very small UK firms, often available only
at a higher level of aggregation.
176 Chapter 5

four-digit or the five-digit level of aggregation, although these were


often modified to exclude secondary products or to ensure com-
parability across various data sources and over time. In a few cases it
was also necessary to split a category or merge two categories into
one in order to obtain an appropriate market definition or avoid us-
ing an industry with very small sales.
The main factors restricting the size of the samples for the adver-
tising regressions were (1) the very limited availability of advertising
data for producer-good industries, (2) the difficulties encountered in
matching advertising to sales figures, and (3) the exclusion of indus-
tries with ambiguous state of competition. In addition, a few observa-
tions were excluded because of government regulation of advertising
or competition in particular industries over certain periods.
The advertising figures used are for display advertising in the
press and on TV by UK manufacturers and importers. They do not
include direct selling expenses or sales promotion expenditure. In a
few industries, advertising of manufacturers’ brands by accredited
dealers, which is often carried out in conjunction with the manufac-
turer, is also included, although it is only a small fraction of the total.
However, I excluded any independent advertising by retailers.
The main problems with the Legion/MEAL figures are (1) the in-
complete coverage of press advertising, which causes a downward
bias, and (2) the fact that the numbers are estimated by pricing the
measured amount of advertising according to the published rates
(i.e., without taking discounts into account), which results in an up-
ward bias for TV expenditure in particular. Both biases may vary
across industries and years. In the case of press advertising, an ad-
ditional complication is that the coverage improved during the 1950s
and the 1960s and it differs slightly between Legion and MEAL.
Finally, the Legion/MEAL data do not include the production costs
of advertisements.
To reduce the measurement error from these factors I adjusted the
Legion/MEAL figures on the basis of data on aggregate annual
expenditure by type of advertising published by the Advertising
Association in Advertising Expenditure 1960 (London: Advertising
Association, 1962) and, subsequently, in Advertising Quarterly. First,
the TV advertising figures were adjusted using estimates of the
average discount across all industries for each year; these were on the
order of 10–17% of the published rates, depending on the year. Sec-
Price Competition, Advertising, and Market Structure 177

ond, all press advertising figures were adjusted upward by 10% to


account for incomplete coverage of press advertising.15 Third, to
ensure comparability of the MEAL 1973 figures with the Legion
1954–1968 data, I adjusted the MEAL figures proportionally, on the
basis of a comparison of Legion with MEAL data for 1968 (a year
for which data from both sources are available). And finally, I
adjusted the Legion/MEAL figures to take into account produc-
tion costs of press and TV advertisements. These were remarkably
stable throughout the period as a fraction of press space or TV time
charges, respectively: about 10% of aggregate expenditure. Hence I
applied a uniform 10% adjustment across industries. Advertising
in other media, such as posters or cinema, was ignored, since cross-
industry data were not available and a uniform adjustment across
industries would be inappropriate (these expenditures differ consid-
erably across industries). Note, however, that these expenditures
were very small: about 5–10% of total advertising expenditure.
Finally, it is worth pointing out that the sales revenue figures are
totals for the combined retail and nonretail market. Distinguishing
between the two markets might have been more appropriate, given
that advertising is mostly targeted toward the retail market, but
such a distinction was not feasible given the data available. How-
ever, in most industries the fraction of retail in total sales should not
change significantly over a ten-year or a twenty-year period. More-
over, the advertising data generally include trade advertising. De-
spite the somewhat lower coverage of this type of advertising in
Legion/MEAL, it is a considerable fraction of the total in some
industries.

15. The reason for applying a uniform adjustment of 10% to all press advertising fig-
ures is as follows. In the case of press advertising, a large part of the difference be-
tween Legion/MEAL and Advertising Association data relates to local advertising by
very small firms and to industrial or retail advertising or advertising of services, all of
which were not relevant for my present purposes. Also, much of the improvement in
Legion press coverage during the 1950s and the 1960s was primarily in these sectors.
In fact, I assumed that the improvement was wholly in these sectors, which in turn
implied that the extent of underreporting of media advertising by larger firms in con-
sumer-good industries was fairly uniform over the period. Hence I applied a uniform
upward adjustment to all Legion press figures. The choice of the 10% figure was based
on a comparison of Legion with Advertising Association aggregate figures for 1968:
the Legion press aggregate was 84% of the Advertising Association aggregate, and
much (but not all) of this must reflect incomplete coverage of press advertising by
larger firms in consumer-good industries.
178 Chapter 5

Table 5.1
Changes in C5 and ADS, 1958–1968
DC5 > 0 DC5 > 0 DC5 < 0 DC5 < 0
D ADS > 0 D ADS < 0 D ADS > 0 D ADS < 0 Total
No. of industries with 1 4 0 1 6
average ADS > 1%
and CHANGE ¼ 1
No. of industries with 7 8 2 6 23
average ADS > 1%
and CHANGE ¼ 0
No. of industries with 1 4 0 0 5
average ADS > 2%
and CHANGE ¼ 1
No. of industries with 4 4 2 6 16
average ADS > 2%
and CHANGE ¼ 0
Notes: The figures in the first two rows are based on twenty-nine advertising-intensive
industries with available comparable data for C5 and ADS for both 1958 and 1968. The
figures in the last two rows are based on twenty-one high-advertising industries with
available comparable data for C5 and ADS for both 1958 and 1968.

5.5 Econometric Models and Results

The empirical analysis of the effects of competition in advertising-


intensive industries in this section is based on a comparison of those
industries affected by the 1956 Act with a control group of industries
not affected. Some information on the joint evolution of concentra-
tion and advertising intensity in the two groups between 1958 and
1968 is given in table 5.1. The figures are based on industries with
unambiguous state of competition and available data for the five-
firm sales concentration ratio, C5, and the advertising-sales ratio,
ADS, for both 1958 and 1968. C5 is the only available concentration
measure at the four-digit level of aggregation, and ADS is used in-
stead of advertising expenditure because of the need to control for
changes in sales during this period. Note that although the theoreti-
cal results of sections 5.2 and 5.3 are cast in terms of both the number
of firms and the concentration ratio, the number of firms is not a very
appropriate measure of market structure in advertising-intensive
industries in practice, since it is too sensitive to the number of small
firms that advertise little or not at all and exist alongside the bigger
advertisers in practically all of these industries. On the other hand,
small nonadvertising firms have little effect on an industry’s total
Price Competition, Advertising, and Market Structure 179

advertising expenditure or intensity or on concentration, which also


implies that C5 is an appropriate measure for our present purposes.
The sample of industries for table 5.1 is rather small, mainly be-
cause of difficulties in matching the industry categories used in the
concentration statistics with those used in the advertising statistics.
One small discrepancy remains: the concentration ratios are for UK
firms, while the advertising-sales ratios are for the UK market. This
does not seriously affect the comparability of the figures, however,
because few industries in this table had significant imports or ex-
ports in 1958 or in 1968.16
It can be seen that there are no industries with a change of com-
petition regime in table 5.1 that experienced a fall in concentration
and a rise in advertising intensity between 1958 and 1968. In con-
trast, all the other outcomes are represented by one or more indus-
tries in the table. Note, however, that within the group of industries
without a change of regime there are only two for which C5 declined
and ADS increased. Hence, while the figures in table 5.1 are consis-
tent with proposition 5.1, they must be interpreted with caution,
since the overall trend in the 1960s appears to have been a rise in
concentration and a fall in advertising intensity.

Econometric Methodology

Clearly, the intensification of price competition following the 1956


legislation was one of several factors that shaped the evolution of
concentration and advertising intensity across UK industries between
the mid-1950s and the mid-1970s. To assess the impact of the 1956
Act across industries, it is therefore necessary to use econometric
analysis. Under this approach, the theoretical predictions regarding
the effect of price competition on the joint evolution of market
structure and advertising are tested only on average. The theoretical
analysis in sections 5.2 and 5.3 suggests estimating the following
reduced-form models:
Concentration ¼ CðS; f ; h; e; tÞ;

Advertising ¼ AðS; f ; h; e; tÞ;

16. Even in those cases, the comparability of the figures would be affected only if ad-
vertising intensity differed considerably between UK products and imported goods or
export intensity varied considerably across leading UK firms in each industry.
180 Chapter 5

where the variables are as defined in section 5.2. Time-invariant


industry-specific characteristics may also influence advertising or
concentration, and their effect will be captured in panel regressions
by the industry-specific effects.
Since the two equations include the same explanatory variables,
there is no efficiency gain from estimating them jointly. Moreover,
the sample would have to be substantially smaller for a joint esti-
mation than for separate regressions, for two reasons. First, it is often
difficult to match the industry definitions used in the concentration
statistics with those used in the advertising statistics. Second, data
for particular industries and years are often not available for one or
the other of the two endogenous variables. For instance, concentra-
tion data are not available for 1954 and 1973, and advertising-sales
ratios for 1975 are somewhat less reliable because of the impact of
the recession of the mid-1970s on advertising expenditure. Thus
the two models will be estimated separately below. Because of the
differences in industry definitions and coverage across statistical
sources, different samples of industries will be used for the two sets
of regressions.
It may also be asked, as in chapter 4, whether it is appropriate
to estimate a regression line or a lower bound for the concentration
equation. But once again there are problems, practical and concep-
tual, with estimating a lower bound. The least squares residuals from
the concentration regressions are negatively skewed, due to the fact
that some industries experienced large increases in concentration
during the period examined. More generally, the estimation of a
lower bound may be problematic in the present case because of the
panel structure of the data and the fact that most of the crucial in-
formation would be lost in a pooled time-series cross section. Thus I
will proceed to test the theory using standard least squares regres-
sions, on the assumption that the theoretical predictions derived for
the benchmark case of symmetric single-product firms carry over to
more complex settings.
Of the variables included in the above reduced-form models, the
degree of horizontal product differentiation h is difficult to measure,
but it should not change significantly over a period of ten or twenty
years in the large majority of industries. The same can be said for
some of the factors that determine advertising effectiveness e and are
related to product characteristics. Both these and h can therefore be
thought of as time-invariant industry effects.
Price Competition, Advertising, and Market Structure 181

TV Advertising and Changes in Advertising Effectiveness

Not all factors affecting advertising effectiveness in the present case


can be thought of as time-invariant industry characteristics, how-
ever. An important development, which was associated with an
exogenous change in advertising effectiveness, was the introduction
of commercial television in the UK in 1955. Since some products
are more suitable for TV advertising than others, this substantially
increased the effectiveness of advertising, and hence advertising
expenditure, in a number of industries in the late 1950s and early
1960s, while in other industries the impact was much smaller. The
overall impact was considerable: aggregate expenditure on display
(as opposed to classified) media advertising as a percentage of GDP
increased from 0.81 in 1956 to 1.08 in 1960, according to the statistics
published by the Advertising Association, and most of this increase
can be attributed to TV advertising.
To control for this factor, a proxy for the effectiveness of TV adver-
tising must be defined. For the advertising regressions, two proxies
will be used. The first is the fraction of TV advertising in total dis-
play advertising for each industry-year pair in the sample (TVTOT ).
Total display advertising is here defined as the sum of press and TV
advertising, that is, other media are ignored because of data limita-
tions. Note, however, that press and TV consistently accounted for
some 90–95% of aggregate display advertising throughout the period.
The theoretical argument for TVTOT can be seen by drawing a dia-
gram with TV advertising on the horizontal axis and press advertis-
ing on the vertical axis to analyze the allocation of a fixed advertising
budget between the two media. In industries where TV advertising is
effective, the ‘‘indifference curves’’ between the two media will be
steep, while the reverse will be the case for industries where TV ad-
vertising is not effective. Starting from a point on the vertical axis
(representing zero TV advertising), an industry will move farther
down the ‘‘isocost’’ curve (toward more TV in total advertising) the
steeper the indifference curves (i.e., the higher the effectiveness of TV
advertising).
While TVTOT adequately captures the overall effect of the intro-
duction of commercial television, including the fact that TV adver-
tising expanded gradually over a period of several years in a manner
that was not uniform across industries, it may be endogenous. The
second proxy, TVINT, is a compromise between the need to con-
182 Chapter 5

trol for the effect of TV advertising and the problem of potential


endogeneity.
TVINT was constructed as follows. All industries were classified
into four groups, according to the fraction of TV in total display ad-
vertising in the late 1960s, that is, after the effect of the introduction
of commercial television had been largely realized: less than 10%,
between 10% and 40%, between 40% and 70%, and higher than 70%.
These cutoff points were chosen on the basis of a histogram of the
fraction of TV in total advertising across all industries in the late
1960s.17 The four groups were assigned the numbers 0, 0.25, 0.55,
and 0.85 (i.e., corresponding to the midpoints of each class except for
the first class, which comprised many industries without any TV
advertising). This number was therefore an index of the equilibrium
level of TV advertising effectiveness in each particular industry.
TVINT was then constructed by multiplying this index by the frac-
tion of aggregate TV advertising in aggregate total advertising for each
year in the sample, in order to account for the gradual expansion of
TV advertising.
As expected, TVINT is strongly correlated with TVTOT, with a
correlation coefficient of 0.91 for the full sample of industries used in
the advertising regressions. On the other hand, since TVINT depends
only on the long-run equilibrium TV advertising effectiveness in any
given industry (which is largely a function of exogenous industry
characteristics) and the time path of the expansion of aggregate
TV advertising, it should not be endogenous with respect to the
advertising-sales ratio of each industry in each particular year.
For the concentration regressions, data limitations resulting from
differences in the industry definitions used in the various statistics
mean that advertising expenditures can often be estimated only ap-
proximately. Again, two proxies for advertising effectiveness will be
used. The first is TVINT. The second is TVCLASS, a variable taking
the value 0, 0.25, 0.55, or 0.85 according to whether the fraction of TV
in total media advertising expenditure for each industry and year was
approximately lower than 10%, between 10% and 40%, between 40%
and 70%, or higher than 70%, respectively. Since TVCLASS can affect
concentration only indirectly, endogeneity is probably not a serious

17. In a few cases where there was evidence of a substantial change of this fraction
after 1968 or of an unstable allocation between press and TV advertising in general, the
average or typical value for the late 1960s and early 1970s was used instead of the 1968
figure.
Price Competition, Advertising, and Market Structure 183

concern in this case. On the other hand, TVCLASS is less likely than
TVINT to be subject to measurement error with respect to the theo-
retical variable e, and this is particularly important in the concentra-
tion regressions because the variation in e in the sample used in these
regressions is relatively low.18

Other Influences on Advertising Expenditure

In both the concentration and the advertising regressions, time dum-


mies will be included among the regressors to control for factors that
may have affected these variables in a more or less uniform way
across industries during the period examined. Several factors that are
thought to have contributed to the overall rise in concentration in the
UK during the 1960s were mentioned in chapter 4.
Regarding advertising, there were important changes during the
1960s in the distribution of manufactured goods: the abolition of re-
sale price maintenance, the growth of large and powerful retailers, the
increased significance of sales promotion,19 and the rise of retail-
ers’ own brands (see Pickering 1974, NBPI 1971, Ward 1973). These
changes reduced the relative importance of brand advertising, for two
reasons. First, more manufacturers were now eager to produce goods
under a retailer’s label. Second, consumers may have become more
sensitive to prices and product availability in large stores and less re-
sponsive to brand image, thus causing manufacturers of advertised
brands to become more anxious to make deals with large retailers and
less keen to advertise. The changes in distribution may also have put
some pressure on manufacturers’ margins.
These changes are thought to have contributed to the substantial
fall in advertising intensity across UK manufacturing in the late
1960s and early 1970s: aggregate expenditure on display advertising
as a percentage of GNP fell from 1.04 in 1963 to 0.93 in 1968 and to
0.84 in 1971, before rising to 0.91 in 1973, the last year before the oil

18. The reason for this is twofold. First, many industries in the samples used in the
concentration regressions, especially in the sample of high-advertising industries, have
a high fraction of TV in total advertising. Second, the first year in these samples is
1958, when TV advertising was already quite established: about 25% of total adver-
tising in 1958 was TV advertising, as compared to zero in 1954 and about 36% in the
late 1960s and early 1970s.
19. Unlike media advertising, promotional activities such as gifts and coupons to
consumers, and financial inducements to distributors affect price competition rather
than increasing perceived quality or providing information about a product.
184 Chapter 5

crisis. Note that time dummies can control for these effects only im-
perfectly, since the impact of changes in distribution patterns on ad-
vertising intensity must have varied across industries. This problem
is relevant mostly for the last two years of the panel used for the
advertising regressions, 1968 and 1973. Other factors that may be
associated with the fall in aggregate advertising intensity in the late
1960s and early 1970s include the operation of prices and incomes
policy, the acceleration of the inflation rate, and the increase in import
penetration.
An additional source of noise in the data, this time during 1954–
1958, is the fact that in many respects British industry did not fully
emerge from the implications of wartime controls until the second
half of the 1950s. Advertising was still low in 1954 compared with
the late 1930s, and it was growing rapidly during the mid-1950s even
before the advent of commercial TV. This evolution can be captured
only partly by a year dummy for 1954, since it was not uniform
across industries.

Price Competition and Concentration: The Samples and Descriptive


Statistics

Two samples of industries are used for the concentration regressions


reported in this chapter. The basic sample includes all industries
with an average or typical advertising-sales ratio of more than 1%
over the relevant period. Since, however, I also wished to focus more
closely on a group of industries for which advertising is clearly a key
strategic variable, I have defined, in addition, a subsample of indus-
tries with average or typical ADS higher than 2% over the relevant
period.
To construct the first of these samples, I started by excluding from
the set of industries with available concentration data all industries
with average or typical advertising-sales ratio lower than 1%. Then I
excluded industries with ambiguous state of competition in 1958 (or,
in a few cases, ambiguous state of competition in the 1960s and early
1970s), as well as industries with a switch of competition regime but
for which concentration data were not available for at least the three
core years 1958, 1963, and 1968. However, I did not exclude indus-
tries without a change in regime and with available data for only a
subset of the period 1958–1968. This left an unbalanced sample of
fifty-six industries, including ten with cancelled agreements, and 204
Price Competition, Advertising, and Market Structure 185

Table 5.2
The samples used in the concentration regressions: advertising-intensive industries
and high-advertising industries
No. of No. of No. of No. of
indus- indus- indus- indus-
tries with tries with tries with tries with
CHANGE ¼ 1 CHANGE ¼ 0 CHANGE ¼ 1 CHANGE ¼ 0
and average and average and average and average
Data available for ADS > 1% ADS > 1% ADS > 2% ADS > 2%
All four years 10 30 5 16
1958, 1963, 1968 0 1 0 1
1963, 1968, 1975 0 11 0 5
1958, 1963 0 3 0 0
1963, 1968 0 1 0 1
Total 10 46 5 23

observations. In most cases observations were available for all four


years in the sample (i.e., 1958, 1963, 1968, and 1975). In some cases,
observations were available for only three or two years.
To construct the subsample of high-advertising industries, I then
simply dropped all industries with average or typical ADS < 2%.
This left twenty-eight industries, including five with canceled agree-
ments, and 104 observations. Table 5.2 gives details on the structure
of these samples, distinguishing between industries with a change
in competition regime after 1958 and industries without such a
change.20 The full set of industries and information on key variables
are given in appendix B (table B2).
Some descriptive statistics on initial levels and changes in concen-
tration are reported in tables 5.3 and 5.4. Table 5.3 reports descriptive
statistics for the concentration ratio, C5, in 1958 for industries with
and industries without a change in competition regime after 1958.
There is clearly some difference between the two groups for indus-
tries with typical ADS > 1%, but this essentially disappears when we
turn to industries with typical ADS > 2%. For this latter class of
industries the mean C5 is 0.707 for the five industries in the sample
with CHANGE ¼ 1 and 0.686 for the seventeen industries with

20. I have also run regressions using more balanced samples, including only indus-
tries with a full set of observations for at least the core period 1958–1968. The results
were very similar to those reported below. Recall that the same was true for exogenous
sunk cost industries. See Symeonidis (2000a) for a discussion of these results (with the
minor qualification that data for 1977, instead of 1975, are used in that paper).
186 Chapter 5

Table 5.3
Concentration in 1958 and competition regime in advertising-intensive and high-
advertising industries
Mean C5 in 1958
(st. deviation of C5)
Advertising-intensive industries
Industries with change of competition regime after 0.737 (0.131)
1958 (n ¼ 10)
Industries without change of competition regime after 0.609 (0.270)
1958 (n ¼ 34)
High-advertising industries
Industries with change of competition regime after 0.707 (0.158)
1958 (n ¼ 5)
Industries without change of competition regime after 0.686 (0.231)
1958 (n ¼ 17)
Notes: The figures are based on industries with available data for 1958; n indicates the
number of industries.

Table 5.4
Average change in C5, 1958–1968 and 1958–1975, in advertising-intensive and high-
advertising industries
DC5 1958–1968 DC5 1958–1975

Advertising-intensive industries
Industries with CHANGE ¼ 1 0.100 0.130
(0.075) (0.101)
n ¼ 10 n ¼ 10
Industries with CHANGE ¼ 0 0.049 0.038
(0.122) (0.139)
n ¼ 31 n ¼ 30
High-advertising industries
Industries with CHANGE ¼ 1 0.136 0.150
(0.082) (0.123)
n¼5 n¼5
Industries with CHANGE ¼ 0 0.047 0.031
(0.098) (0.113)
n ¼ 17 n ¼ 16

Notes: The figures in the first column are based on industries with available observa-
tions for both 1958 and 1968. The figures in the second column are based on industries
with available observations for both 1958 and 1975. The figures in parentheses are
standard deviations. n indicates the number of industries.
Price Competition, Advertising, and Market Structure 187

CHANGE ¼ 0 and available data for 1958. As will be seen below, this
class is the crucial one for testing the theoretical predictions devel-
oped in this chapter.
On the other hand, table 5.4 shows that the average increase in C5
after 1958 was much larger for industries affected by the legislation
than for industries not affected, whatever the sample used. For in-
stance, for the ADS > 2% sample, the average change in C5 between
1958 and 1968 was 13.6 percentage points for industries with a
change in competition regime and only 4.7 percentage points for the
control group. For the ADS > 1% sample, the respective numbers for
the average change in C5 over 1958–1968 are 10.0 percentage points
and 4.9 percentage points. Similar differences are observed for the
1958–1975 period. These differences are so large that they can be
expected to persist even when controlling for other variables.
A comparison between industries that experienced a large increase
in C5 over the period 1958–1975 and industries that experienced a
large decrease in C5 over the same period is also very informative.
The comparison is limited to the forty advertising-intensive industries
with available data for both these years. Of the five industries with the
largest rise in C5, two (biscuits for human consumption; washing
machines, electrically operated) are industries with CHANGE ¼ 1,
and three (corsets and brassieres; television receiving sets; knitted,
netted, or crocheted goods: hosiery) have CHANGE ¼ 0. However,
all five industries with the largest fall in C5 (tufted carpets, carpeting,
and carpet floor rugs; vegetables, quick frozen; fish and fish prod-
ucts, quick frozen; cutlery; powered industrial trucks and industrial
tractors) have CHANGE ¼ 0.

Price Competition and Concentration: Econometric Model and


Results

The basic econometric model for concentration in this chapter is

Concit ¼ a i þ b 1 ln Salesit þ b 2 lnðK=xÞit þ b 3 TVCLASSit þ b4 Y63


þ b 5 Y68 þ b6 Y75 þ b7 CHANGE  Y63 þ b8 CHANGE  Y68

þ b 9 CHANGE  Y75 þ uit ;


where ‘‘conc’’ is either the four-digit industry five-firm sales con-
centration ratio, C5, or its logistic transformation, logit C5 ¼ ln½C5=
ð1  C5Þ; ‘‘sales’’ is the four-digit industry sales revenue deflated
188 Chapter 5

either by the general producer price index ðSSÞ or by an industry-


specific producer price index ðDSÞ; ‘‘K/x’’ is either the three-digit
industry capital stock of the average plant, K=N, or the three-digit
industry capital-labor ratio, K=L; TVCLASS is as defined above; Y63,
Y68, and Y75 are time dummies for 1963, 1968, and 1975, respec-
tively; and the interaction terms should capture any differences in
the evolution of concentration after 1958 between industries with a
change in competition regime ðCHANGE ¼ 1Þ and industries without
such a change ðCHANGE ¼ 0Þ. Thus the coefficient on CHANGE
 Y63 measures the effect of the 1956 Act on concentration by 1963,
the coefficient on CHANGE  Y68 measures the effect by 1968, and
the coefficient on CHANGE  Y75 measures the effect by 1975. The
benchmark year is 1958. Note that the arguments put forward in
chapter 4 regarding the potential endogeneity of some of the inde-
pendent variables are valid here as well.
Table 5.5 contains results for industries with typical ADS > 1%,
including ten industries with canceled agreements, using a fixed-
effects specification.21 The reported standard errors are heteroskeda-
sticity-consistent, adjusted for small sample bias following MacKinnon
and White (1985). The coefficients on CHANGE  Y75 are everywhere
positive and statistically significant at the 5% or the 10% level, thus
providing evidence of a positive competition-concentration relation-
ship in advertising-intensive industries. According to these estimates,
the effect of the cartel legislation on market structure in advertising-
intensive industries was a rise in the five-firm concentration ratio of
about six percentage points. Moreover, the results suggest that this
effect was only partly realized by 1963 or even 1968. Recall that in
chapter 4 some effect after 1968 was also found for exogenous sunk
cost industries, but it was, in that case, quite smaller—and could be
simply due to time lags in the adjustment of concentration to its new
long-run equilibrium. Such an interpretation seems less plausible for
the present sample.
There are several reasons why one should not overemphasize this
difference between exogenous sunk cost and advertising-intensive
industries. First, and perhaps foremost, the 95% confidence intervals
for the coefficients on CHANGE  Y68 or on CHANGE  Y75 in the
two classes of industries largely overlap, so there is no strong evi-

21. The random-effects model gives similar results with respect to the market size
variable and the competition variables, but it is always rejected by the Hausman test
in favor of the fixed-effects model.
Price Competition, Advertising, and Market Structure 189

dence of any difference in the timing of the impact of the 1956 Act on
concentration across classes of industries. Second, a few agreements
in advertising-intensive industries continued until late into the 1960s
before being abandoned. Third, the effect of the Act after 1968 in
advertising-intensive industries appears to be more pronounced in
regressions using logit C5, and it is precisely in these regressions that
heteroskedasticity is also more pronounced, which suggests that
these particular results may be disproportionately influenced by a
few industries with very high concentration. Fourth, the rise in con-
centration in previously cartelized industries after 1968 does not
appear particularly pronounced when one looks at the descriptive
statistics of table 5.4. And, finally, the descriptive statistics in table
5.4, especially when read in conjunction with those in table 4.3, sug-
gest that an important factor in the estimated impact of the 1956 Act
over time was the evolution of industries not affected by the legisla-
tion. In particular, while the average C5 increased over 1968–1975 in
exogenous sunk cost industries without a change in competition re-
gime, it fell slightly in the equivalent group of advertising-intensive
industries.
All this is not to say that any estimated effect of the 1956 Act on
concentration after 1968 in advertising-intensive industries is neces-
sarily more apparent than real. One potentially relevant factor in this
respect is the widespread use of resale price maintenance in these
industries, in contrast with exogenous sunk cost industries. It may
be the case that resale price maintenance delayed or hindered the
emergence of fully effective price competition among manufacturers
in some previously cartelized industries. Its abolition in the second
half of the 1960s may therefore be one factor behind the rise in con-
centration after 1968 in these industries.
Another potential factor is the evolution of advertising. A fall in
advertising expenditure in previously cartelized industries may have
occurred and helped absorb part of the pressure on net profits for
several years after 1958. Once advertising had fallen to very low
levels, however, it could not fall any farther, which may be why we
see an effect of the Act on concentration continuing after 1968 in
some advertising-intensive industries (assuming that competition
further intensified after that date).
A final interpretation is that the changes in the pattern of distri-
bution and the consequent fall in advertising across manufacturing
industries after the mid-1960s created favorable circumstances for a
Table 5.5

190
Regression results for concentration for industries with average ADS > 1% (fixed-effects estimation)
Dependent variable: C5 Dependent variable: logit C5
ln SS 0.102 — 0.102 — 0.71 — 0.73 —
(0.018) (0.018) (0.12) (0.13)
ln DS — 0.085 — 0.085 — 0.55 — 0.57
(0.022) (0.022) (0.14) (0.15)
ln K=N 0.001 0.004 — — 0.28 0.31 — —
(0.029) (0.031) (0.22) (0.24)
ln K=L — — 0.012 0.011 — — 0.34 0.33
(0.032) (0.035) (0.27) (0.28)
TVCLASS 0.023 0.013 0.024 0.013 0.54 0.44 0.52 0.42
(0.036) (0.039) (0.036) (0.039) (0.26) (0.28) (0.26) (0.28)
Y63 0.050 0.049 0.053 0.050 0.43 0.41 0.42 0.39
(0.020) (0.021) (0.018) (0.019) (0.13) (0.14) (0.12) (0.13)
Y68 0.095 0.094 0.102 0.097 0.81 0.79 0.81 0.76
(0.029) (0.031) (0.027) (0.029) (0.22) (0.23) (0.21) (0.22)
Y75 0.114 0.114 0.125 0.120 0.99 0.97 1.01 0.95
(0.040) (0.044) (0.037) (0.042) (0.28) (0.30) (0.29) (0.31)
CHANGE  Y63 0.025 0.027 0.025 0.027 0.16 0.17 0.16 0.18
(0.031) (0.033) (0.031) (0.032) (0.19) (0.20) (0.19) (0.20)

Chapter 5
CHANGE  Y68 0.039 0.046 0.038 0.046 0.29 0.34 0.29 0.34
(0.032) (0.032) (0.032) (0.032) (0.21) (0.21) (0.21) (0.21)
CHANGE  Y75 0.064 0.065 0.063 0.064 0.67 0.69 0.67 0.69

Price Competition, Advertising, and Market Structure


(0.032) (0.036) (0.032) (0.035) (0.25) (0.28) (0.25) (0.28)
R2 0.36 0.30 0.37 0.30 0.32 0.25 0.32 0.25
R 2LSDV 0.96 0.95 0.96 0.95 0.93 0.93 0.93 0.93
Hausman statistic 33.0 32.0 30.6 17.7 77.9 41.2 51.6 38.1
Prob-value 0.0001 0.0002 0.0003 0.04 A0 A0 A0 A0
No. of industries 56 56 56 56 56 56 56 56
No. of industries 10 10 10 10 10 10 10 10
with CHANGE ¼ 1
No. of observations 204 204 204 204 204 204 204 204
2
Notes: Heteroskedasticity-consistent standard errors in parentheses. R is derived from transforming the data to obtain deviations from in-
dustry means and applying OLS to the transformed data. R 2LSDV is obtained from applying OLS to the untransformed data after including a set
of industry dummies among the regressors.

191
192 Chapter 5

fall in concentration. However, these factors may have had a bigger


impact on industries not affected by the 1956 Act if the fall in adver-
tising expenditure since the late 1960s was more pronounced in these
industries than in previously cartelized ones, a fact which could be
associated with the effect of the 1956 Act on advertising during
1958–1968. Note that most of these stories involve some analysis of
the joint evolution of advertising and market structure, so it is prob-
ably best to return to these issues after examining the econometric
results on advertising later in this chapter.
A second result from table 5.5 is the negative, large, and statisti-
cally significant coefficients on the market size variables, which do
not differ much between regressions using ln SS and those using
ln DS. To check whether this may be related to the downward trend
in advertising expenditure in UK manufacturing during the late
1960s and early 1970s, I reestimated the model for the period 1958–
1968 only. The market size coefficients had almost the same magnitude
and again were statistically significant. Finally, note that replacing
TVCLASS with TVINT or dropping this variable altogether does not
significantly affect the results.

Price Competition and Concentration: The ADS > 2% Sample

The evidence reported in table 5.5 suggests that the long-run effect of
competition and market size on concentration is not significantly
different in advertising-intensive industries than in exogenous sunk
cost industries. This may be because the 1% cutoff point is too low. In
particular, the results in table 5.5 may be driven by industries with
medium advertising intensity, and in these industries a positive
competition effect on concentration may be due to the fact that ad-
vertising cannot fall by much anyway. To obtain more decisive evi-
dence on the present theory, it is therefore necessary to examine the
sample of industries with typical ADS > 2%, that is, industries where
endogenous sunk costs represent a more significant fraction of total
sunk costs.
Results for the ADS > 2% sample using a fixed-effects specification
are presented in table 5.6, and results for the random-effects model,
using generalised least squares estimation, are reported in table 5.7.
The Hausman test is inconclusive in this instance. In any case, the
two sets of results are similar, except for the coefficients on the setup
cost proxies, which are statistically significant in the random-effects
model only.
Price Competition, Advertising, and Market Structure 193

The first thing to note is that the market size coefficients in these
regressions are nowhere statistically significant, suggesting that the
concentration-market size relationship breaks down in industries
with high advertising intensity. This is consistent with Sutton’s (1991)
robust prediction that concentration may either rise or fall as market
size increases in industries with significant endogenous sunk costs. 22
On the other hand, the competition effect is statistically significant
and large, although its magnitude should be interpreted with caution
since the number of industries with a change of competition regime
in the ADS > 2% sample is only five (and none of these has a very
high typical ADS).23 In any case, the 95% confidence intervals for
the coefficients on CHANGE  Y68 or on CHANGE  Y75 in these and
the previous regression results largely overlap, so there is no strong
evidence of any difference in the magnitude of the long-run competi-
tion effect on concentration across classes of industries.
To check whether any of the five industries with a change of re-
gime has a disproportionate influence on the results, I reestimated
the model, excluding each of the five industries in turn. The coeffi-
cient on CHANGE  Y75 remained significant at the 5% or the 10%
level in all cases.
Since TVCLASS is a somewhat crude proxy, it may also be inter-
esting to see what happens if it is dropped altogether. In fact, the
results are very similar to those in tables 5.6 and 5.7. Finally, using
TVINT instead of TVCLASS again produces results similar to those
reported here. The absence of any significant effect of TVCLASS or
the market size variables on concentration is not inconsistent with
the theory of section 5.2: a rise in advertising effectiveness e or market
size S causes both advertising expenditure and gross profit to in-
crease, given the initial number of firms, so the effect on concentra-
tion is ambiguous. Only the joint outcome DA  a 0 and DC  b 0 is
ruled out. In fact, the lack of any systematic market size effect on
concentration is one of the predictions of the specific example pre-
sented in section 5.3.
A potential concern with the above results is that none of the
industries with a change in competition regime in the samples used
had an advertising-sales ratio higher than 5% in 1958, and many had

22. Other econometric studies that have confirmed this prediction include Sutton (1991,
chapter 5), Matraves (1992), Lyons and Matraves (1996), and Robinson and Chiang
(1996).
23. The five industries are biscuits, cocoa products, condensed milk, electric cookers
(stoves), and washing machines.
Table 5.6

194
Regression results for concentration for industries with average ADS > 2% (fixed-effects estimation)
Dependent variable: C5 Dependent variable: logit C5
ln SS 0.013 — 0.013 — 0.21 — 0.20 —
(0.028) (0.029) (0.24) (0.25)
ln DS — 0.037 — 0.037 — 0.09 — 0.08
(0.026) (0.026) (0.21) (0.20)
ln K=N 0.002 0.003 — — 0.05 0.01 — —
(0.059) (0.059) (0.32) (0.33)
ln K=L — — 0.013 0.018 — — 0.07 0.07
(0.069) (0.069) (0.43) (0.44)
TVCLASS 0.047 0.057 0.045 0.056 0.32 0.15 0.32 0.15
(0.057) (0.058) (0.057) (0.058) (0.49) (0.48) (0.49) (0.48)
Y63 0.022 0.018 0.026 0.021 0.31 0.26 0.31 0.24
(0.035) (0.035) (0.032) (0.032) (0.22) (0.21) (0.21) (0.21)
Y68 0.034 0.026 0.043 0.034 0.29 0.21 0.29 0.16
(0.056) (0.055) (0.052) (0.051) (0.32) (0.31) (0.33) (0.32)
Y75 0.007 0.007 0.023 0.006 0.20 0.05 0.19 0.02
(0.083) (0.081) (0.082) (0.079) (0.47) (0.45) (0.52) (0.50)
CHANGE  Y63 0.036 0.035 0.034 0.033 0.14 0.13 0.14 0.14
(0.051) (0.051) (0.051) (0.051) (0.28) (0.28) (0.28) (0.28)

Chapter 5
CHANGE  Y68 0.086 0.083 0.083 0.080 0.61 0.61 0.61 0.63
(0.052) (0.052) (0.051) (0.051) (0.32) (0.30) (0.32) (0.31)
CHANGE  Y75 0.124 0.127 0.120 0.124 0.96 1.01 0.96 1.03

Price Competition, Advertising, and Market Structure


(0.059) (0.058) (0.060) (0.058) (0.44) (0.45) (0.47) (0.48)
R2 0.29 0.30 0.29 0.30 0.22 0.22 0.22 0.22
R 2LSDV 0.94 0.94 0.94 0.94 0.93 0.93 0.93 0.93
No. of industries 28 28 28 28 28 28 28 28
No. of industries 5 5 5 5 5 5 5 5
with CHANGE ¼ 1
No. of observations 104 104 104 104 104 104 104 104
2
Notes: Heteroskedasticity-consistent standard errors in parentheses. R is derived from transforming the data to obtain deviations from in-
dustry means and applying OLS to the transformed data. R 2LSDV is obtained from applying OLS to the untransformed data after including a set
of industry dummies among the regressors.

195
Table 5.7

196
Regression results for concentration for industries with average ADS > 2% (random-effects estimation)
Dependent variable: C5 Dependent variable: logit C5
ln SS 0.019 — 0.002 — 0.03 — 0.10 —
(0.023) (0.023) (0.23) (0.24)
ln DS — 0.001 — 0.010 — 0.14 — 0.23
(0.021) (0.020) (0.19) (0.20)
ln K=N 0.067 0.057 — — 0.57 0.49 — —
(0.033) (0.034) (0.26) (0.26)
ln K=L — — 0.083 0.077 — — 0.70 0.66
(0.048) (0.048) (0.34) (0.34)
TVCLASS 0.064 0.072 0.060 0.067 0.04 0.04 0.07 0.001
(0.057) (0.055) (0.059) (0.057) (0.55) (0.52) (0.56) (0.53)
Y63 0.007 0.007 0.006 0.005 0.14 0.12 0.13 0.10
(0.029) (0.029) (0.029) (0.029) (0.20) (0.21) (0.20) (0.21)
Y68 0.005 0.005 0.008 0.010 0.15 0.17 0.17 0.21
(0.035) (0.034) (0.037) (0.037) (0.24) (0.23) (0.25) (0.25)
Y75 0.052 0.053 0.063 0.068 0.53 0.56 0.62 0.69
(0.048) (0.048) (0.055) (0.055) (0.33) (0.32) (0.38) (0.37)
CHANGE  Y63 0.035 0.034 0.041 0.040 0.04 0.03 0.09 0.08
(0.051) (0.051) (0.048) (0.048) (0.32) (0.32) (0.29) (0.28)

Chapter 5
CHANGE  Y68 0.088 0.087 0.094 0.093 0.59 0.56 0.64 0.62
(0.041) (0.042) (0.041) (0.042) (0.29) (0.29) (0.27) (0.28)
CHANGE  Y75 0.125 0.127 0.136 0.137 1.00 1.01 1.09 1.10

Price Competition, Advertising, and Market Structure


(0.044) (0.045) (0.048) (0.048) (0.45) (0.45) (0.46) (0.46)
Constant 0.914 0.725 0.634 0.504 1.95 0.09 0.39 1.74
(0.251) (0.228) (0.247) (0.216) (2.48) (2.03) (2.53) (2.06)
2
R 0.26 0.21 0.27 0.23 0.20 0.17 0.20 0.19
Hausman statistic 21.7 54.2 8.8 10.4 30.2 14.4 — 8.2
Prob-value 0.01 A0 0.46 0.32 0.0004 0.11 0.51
No. of industries 28 28 28 28 28 28 28 28
No. of industries 5 5 5 5 5 5 5 5
with CHANGE ¼ 1
No. of observations 104 104 104 104 104 104 104 104
Note: Heteroskedasticity-consistent standard errors in parentheses.

197
198 Chapter 5

advertising-sales ratios much lower than that. As pointed out in sec-


tion 5.2, the effect of price competition on concentration can be neg-
ative only when there is scope for a substantial fall in advertising
intensity. Advertising-sales ratios of 2–3% may still be too low in
that respect: if the advertising-sales ratio cannot fall by very much
anyway, it cannot offset the negative effect of price competition on
gross profit, so concentration has to rise. Could, then, the estimated
positive competition effect on concentration in advertising-intensive
industries be specific to the sample used here?
The answer is ‘‘probably not.’’ The main reason is that the esti-
mated positive competition effect on concentration is stronger in the
subsample of industries with typical ADS > 2% than in the full
sample of industries with typical ADS > 1%. While this result should
not be overemphasized, given the small number of industries with a
change in competition regime in these samples, it certainly provides
no indication that the competition effect on concentration may be
weakening, on average, as advertising intensity increases. I should
also point out, however, that the econometric results can identify only
overall effects, and do not imply that a negative competition effect on
concentration will never occur in advertising-intensive industries.
This point will be further explored in the context of one of the case
studies presented below. What the econometric evidence seems to
suggest is that a positive competition effect on concentration is the
most likely outcome in advertising-intensive industries.

Price Competition and Advertising Intensity: The Samples and


Descriptive Statistics

I now turn to the impact of price competition on advertising inten-


sity. The basic sample of industries used for the econometric analysis
consists of an unbalanced panel of seventy-one industries with av-
erage ADS > 1% over the three core years 1958, 1963, and 1968.24 A
second sample, including fifty-one industries with average ADS >

24. One reason why low-advertising industries were excluded from the present anal-
ysis is that any systematic effect of price competition on advertising will be more dif-
ficult to identify and also less relevant in industries where advertising is not a key
strategic variable. Moreover, large proportional changes in advertising intensity in
low-advertising industries may have an influence on the results which is dispropor-
tionate to their economic significance. Finally, note that all the theoretical predictions
in this chapter are for advertising-intensive industries.
Price Competition, Advertising, and Market Structure 199

Table 5.8
The samples used in the advertising regressions: advertising-intensive industries and
high-advertising industries
No. of No. of No. of No. of
indus- indus- indus- indus-
tries with tries with tries with tries with
CHANGE ¼ 1 CHANGE ¼ 0 CHANGE ¼ 1 CHANGE ¼ 0
and average and average and average and average
Data available for ADS > 1% ADS > 1% ADS > 2% ADS > 2%
All five years 12 31 8 23
1954, 1958, 1963, 1968 2 9 1 5
1958, 1963, 1968, 1973 1 0 1 0
1954, 1958, 1963 0 2 0 2
1958, 1963, 1968 1 2 1 2
1963, 1968, 1973 0 6 0 5
1958, 1963 0 2 0 2
1963, 1968 0 3 0 1
Total 16 55 11 40

2% over the three core years 1958, 1963, and 1968, was also con-
structed. Note that sample selection is essentially done here on the
basis of the average ADS—which is likely to be largely driven by ad-
vertising effectiveness, an exogenous variable—so there is no selec-
tion bias. The samples do not include industries with ambiguous
state of competition in 1958 (or in the 1960s and early 1970s). Nor do
they include industries with available observations only for 1954–
1958 or 1968–1973. However, they do include industries with avail-
able data for a subset of the period 1958–1968.25 In most cases
observations are available for all five years in the sample (1954, 1958,
1963, 1968 and 1973). The two samples include 306 and 219 obser-
vations, respectively. Table 5.8 gives details on their structure.26 The
full set of industries and information on key variables are given in
table B3 of appendix B.
Tables 5.9 and 5.10 present descriptive statistics on initial levels
and changes in the advertising-sales ratio, ADS, separately for in-

25. Industries with a switch of regime and with advertising or sales data available for
only a subset of the period 1958–1968 would have normally been excluded, but there
were no such cases in these samples.
26. I have also run regressions using more balanced samples, including only indus-
tries with a full set of observations for at least the core period 1958–1968. The results
were very similar to those reported below.
200 Chapter 5

Table 5.9
ADS, 1954 and 1958, and competition regime in advertising-intensive and high-
advertising industries
Mean ADS in 1954 Mean ADS in 1958
(st. deviation of ADS) (st. deviation of ADS)
Advertising-intensive industries
Industries with change of 0.025 (0.023) 0.035 (0.028)
competition regime after
1958 (n ¼ 14)
Industries without change 0.057 (0.067) 0.071 (0.076)
of competition regime
after 1958 (n ¼ 42)
High-advertising industries
Industries with change of 0.032 (0.027) 0.045 (0.032)
competition regime after
1958 (n ¼ 9)
Industries without change 0.076 (0.071) 0.093 (0.080)
of competition regime
after 1958 (n ¼ 30)
Notes: The figures are based on industries with available data for 1954 and 1958;
n indicates the number of industries.

dustries with a change in competition regime after 1958 and indus-


tries without a change in regime. Clearly, there is a large difference in
initial conditions between the two groups. For instance, in table 5.9
the mean ADS in 1958 is 3.5% for fourteen advertising-intensive
industries with CHANGE ¼ 1 and 7.1% for forty-two advertising-
intensive industries with CHANGE ¼ 0 (all industries in the latter
group were noncollusive in the 1950s). The picture is substantially
the same in 1954 and for the subsample of high-advertising indus-
tries. How are we to interpret this negative correlation between col-
lusive pricing and advertising intensity? Recall that this issue was
discussed in a slightly different context in chapter 3, where I argued
in favor of an interpretation that sees the causality as going from
advertising effectiveness (an exogenous variable) to the likelihood of
collusive pricing. Under that interpretation, collusion on price is
more likely to occur in industries with low advertising effectiveness
(and therefore also low advertising intensity).
Table 5.10 shows why an alternative interpretation, namely, that
price collusion reduces advertising intensity, is not valid. This table
compares the evolution of the mean ADS in the two groups over
1958–1968 and also over 1958–1973. In both groups ADS fell, and the
Table 5.10

Price Competition, Advertising, and Market Structure


Percentage change of mean ADS in advertising-intensive and high-advertising industries, 1958–1968 and 1958–1973
Percentage Percentage
Mean ADS Mean ADS change Mean ADS Mean ADS change
1958 1968 1958–1968 1958 1973 1958–1973
Advertising-intensive industries
Industries with CHANGE ¼ 1 0.037 0.023 36.4% 0.037 0.019 47.1%
(0.027) (0.016) (0.029) (0.016)
n ¼ 16 n ¼ 16 n ¼ 13 n ¼ 13
Industries with CHANGE ¼ 0 0.071 0.063 12.3% 0.084 0.050 40.8%
(0.077) (0.069) (0.083) (0.047)
n ¼ 42 n ¼ 42 n ¼ 31 n ¼ 31
High-advertising industries
Industries with CHANGE ¼ 1 0.045 0.029 36.6% 0.045 0.024 47.6%
(0.029) (0.017) (0.032) (0.017)
n ¼ 11 n ¼ 11 n¼9 n¼9
Industries with CHANGE ¼ 0 0.094 0.083 11.6% 0.108 0.065 39.8%
(0.080) (0.071) (0.084) (0.046)
n ¼ 30 n ¼ 30 n ¼ 23 n ¼ 23
Notes: The figures in the first three columns are based on industries with available observations for both 1958 and 1968. The figures in the last
three columns are based on industries with available observations for both 1958 and 1973. The figures in parentheses are standard deviations.
n indicates the number of industries.

201
202 Chapter 5

fall was actually more pronounced in the previously collusive group.


While in 1958 noncollusive industries had an average advertising-
sales ratio about twice as high as that of collusive industries, in the
late 1960s and early 1970s the mean ADS of the control group was
nearly three times as high as the mean ADS of the previously collu-
sive group.27 This is certainly not consistent with the view that price
competition promotes advertising. In fact, the reverse could be true.
Does this difference persist when we control for other variables, and
is it statistically significant? To answer these questions, I now turn to
the econometric analysis.

Price Competition and Advertising Intensity: Econometric


Specification

The econometric model that I will use is a panel data model with
industry-specific effects. The basic specification is
ln ADSit ¼ a i þ g1 ln SSDOMit þ g2 lnðK=LÞit þ g3 TVINTit

þ g4 Y54 þ g5 Y63 þ g6 Y68 þ g7 Y73 þ g8 CHANGE  Y54


þ g9 CHANGE  Y63 þ g10 CHANGE  Y68

þ g11 CHANGE  Y73 þ eit ;


where ADS is the industry advertising-sales ratio, SSDOM is sales
revenue in the UK market deflated by the general producer price
index, TVINT and K=L are as previously defined, and Y54, Y63, Y68,
and Y73 are time dummies for 1954, 1963, 1968, and 1973, respec-
tively. As always, the interaction terms should reveal any differences
in the evolution of advertising intensity after 1958 between industries
with a change in competition regime ðCHANGE ¼ 1Þ and industries
without such a change ðCHANGE ¼ 0Þ. In other words, the coefficient
on CHANGE  Y63 measures the effect of the 1956 legislation on
ln ADS by 1963, the coefficient on CHANGE  Y68 measures the effect
by 1968, and the coefficient on CHANGE  Y73 measures the effect

27. Moreover, of the five industries with the largest percentage rise in ADS between
1958 and 1968, only one (chocolate confectionery) has CHANGE ¼ 1, while four (razors
and blades; sanitary towels; ice cream and ice lollies (ice pops); margarine) have
CHANGE ¼ 0. On the other hand, three out of five industries with the largest percent-
age fall in ADS between 1958 and 1968 have CHANGE ¼ 1 (typewriters; domestic
refrigerators and deep freezers; lead pencils), and only two (baby carriages; toilet
brushes) have CHANGE ¼ 0.
Price Competition, Advertising, and Market Structure 203

by 1973. The benchmark year is 1958. Moreover, the coefficient on


CHANGE  Y54 measures whether the evolution of advertising in-
tensity was in any way different between collusive and noncollusive
industries during 1954–1958 (i.e., before the implementation of the
legislation). Finally, note that CHANGE is defined here according to
the industry categories used for ADS, that is, the level of aggregation
is sometimes the four-digit and sometimes the five-digit industry
level.
The reason for using a log transformation of ADS as the dependent
variable is to avoid the difficulties that arise from the fact that ADS
is bounded below by zero. Preliminary tests revealed a marked ten-
dency for the residual variances to increase with fitted values in re-
gressions with the untransformed advertising-sales ratio. In addition
to heteroskedasticity, the coefficients on the competition variables in
these regressions may be biased, given that cartelized industries had,
on average, a much lower ADS than noncartelized industries, both in
1958 and throughout the period. In particular, if the general trend
during the period under study involved larger absolute changes in
advertising intensity in industries with higher advertising than in
those with lower advertising, as seems to be the case, then using the
untransformed ADS as dependent variable might result in an under-
estimation of any changes in ADS in previously collusive industries.
An implication of using a log transformation of the dependent
variable and of including ln SSDOM and time dummies among the
regressors is that the theoretical distinction between the impact of
competition on advertising expenditure and on advertising intensity
is blurred in practice: the choice of the dependent variable (advertis-
ing expenditure or intensity) does not affect the coefficients and
standard errors on the interaction terms or TVINT.
As usual, the potential endogeneity of CHANGE may be a cause
for concern. At first sight, this could be more serious here than in the
concentration regressions because of the difference in initial con-
ditions between collusive and noncollusive industries. However, as
shown in tables 5.9 and 5.10, cartelization in the 1950s is negatively
correlated with advertising intensity throughout the period, not just for
1958 or 1954. This suggests that although collusion seems to be cor-
related with variables that affect advertising intensity and are not
included in the model, these variables are time-invariant, industry-
specific characteristics. Any such unspecified variables will be cap-
tured in the present model by the industry effects, and there will be
204 Chapter 5

no endogeneity bias, provided an estimation method that accounts


for the correlation between regressors and industry effects is used.
But this does not fully settle the issue. There could be some unob-
served characteristic that differs between the two groups and causes
them to evolve in different ways after 1958; then one might wrongly
attribute this evolution to the effect of the legislation. For instance,
suppose that consumers’ experience with a product is inversely re-
lated to advertising effectiveness, so that newer products are asso-
ciated with a higher level of advertising, everything else being equal.
Suppose also that newer products are less likely to be subject to
collusion. If both these conjectures are true, then the decrease in ad-
vertising expenditure after 1958 in previously collusive British man-
ufacturing industries could be partly due to the fact that the products
involved were more mature than the products produced in non-
collusive industries.
An indirect but very informative check of whether this type of bias
is likely to be present in the data is provided by a comparison of
changes in ADS in the two groups before the effective implementation
of the legislation (i.e., during 1954–1958). Table 5.9 shows that the
mean ADS for fourteen advertising-intensive industries with collu-
sive agreements, and subsequently a change in competition regime,
was 2.5% in 1954 and 3.5% in 1958, which implies a 40% increase in
the mean ADS during this period. For forty-two industries in the
control group, on the other hand, the mean ADS was 5.7% in 1954
and 7.1% in 1958, which corresponds to a 25% increase. Thus ADS
appears to have been rising somewhat more rapidly, in proportional
terms, in collusive advertising-intensive industries than in noncol-
lusive ones during 1954–1958.28
Did this pattern continue after 1958? The answer is no. As shown
in table 5.10, the mean ADS for sixteen advertising-intensive indus-
tries with a change in competition regime declined from 3.7% in 1958
to 2.3% in 1968, which represents a 36.4% fall. At the same time, the
mean ADS for forty-two advertising-intensive industries in the con-
trol group fell from 7.1% in 1958 to 6.3% in 1968, a 12.3% fall. In other
words, ADS fell by much more, in proportional terms, in advertising-
intensive industries with a change in regime than in advertising-

28. The reasons for looking here at proportional rather than absolute changes in ADS
are the same as the reasons for using ln ADS rather than the untransformed ADS as
dependent variable in the regressions.
Price Competition, Advertising, and Market Structure 205

intensive industries without such a change during 1958–1968. This is


very different from what was happening before 1958, and so it can-
not be attributed to any underlying trend in ADS that differed be-
tween the two groups.

Price Competition and Advertising Intensity: Econometric Results

Tables 5.11 and 5.12 contain regression results for the two samples.
Since industries with a change in competition regime had, on aver-
age, a much lower advertising-sales ratio throughout the period than
industries in the control group, the competition variable CHANGE
is probably correlated with the industry-specific effects in the panel
data model described above. To obtain consistent estimates, fixed-
effects estimation was used throughout.
Preliminary tests suggested the presence of a clear outlier, namely, the
1963 observation for the washing machine industry. The advertising-
sales ratio reached a record 12.6% in 1963 in this industry, compared
with 4.4% in 1958 and 2.9% in 1968. As described in detail in the
following section, this was due to very special circumstances in that
industry, namely, the temporary entry of firms that used press ad-
vertising as part of their process of distribution. These special cir-
cumstances and the fact that this type of advertising was definitely
nonstandard justify dropping this observation from the sample. In
any case, this affects only the coefficient on CHANGE  Y63, which is
smaller in absolute value and statistical significance when the obser-
vation is included; the other results are not significantly affected.
The two tables report results for the basic model as well as for
several alternative specifications. Thus, I report regression results
using TVINT, and also using TVTOT. Recall that the main concern
with TVTOT is its potential endogeneity. However, this may not be a
serious problem after all: in preliminary regressions using TVINT as
an instrument for TVTOT, the Wu-Hausman test always rejected the
null hypothesis of a significant effect on the estimated coefficients of
the potential endogeneity of TVTOT even at the 20% significance
level. Moreover, since both TVINT and TVTOT are at best imperfect
controls for the effect of TV advertising, I also report results for a
specification that does not include either of these variables. Finally,
I have also experimented with a specification including the interac-
tion variables ln SSDOM  Y54, ln SSDOM  Y63, ln SSDOM  Y68,
and ln SSDOM  Y73 in order to check whether the negative link
Table 5.11
Regression results for advertising intensity for industries with average ADS > 1%
(fixed-effects estimation)
Dependent variable: ln ADS
ln SSDOM 0.21 0.33 0.22 0.34 0.21 0.34
(0.10) (0.10) (0.10) (0.10) (0.11) (0.11)
ln SSDOM  Y54 — 0.05 — 0.06 — 0.06
(0.06) (0.06) (0.07)
ln SSDOM  Y63 — 0.10 — 0.09 — 0.11
(0.05) (0.05) (0.05)
ln SSDOM  Y68 — 0.17 — 0.17 — 0.19
(0.05) (0.05) (0.05)
ln SSDOM  Y73 — 0.21 — 0.21 — 0.22
(0.06) (0.06) (0.06)
ln K=L 0.11 0.13 0.04 0.19 0.01 0.25
(0.18) (0.18) (0.17) (0.17) (0.17) (0.17)
TVINT 2.02 1.73 — — — —
(0.69) (0.64)
TVTOT — — 0.62 0.59 — —
(0.16) (0.15)
Y54 0.09 0.34 0.13 0.43 0.35 0.20
(0.12) (0.63) (0.10) (0.62) (0.10) (0.66)
Y63 0.04 0.99 0.09 0.90 0.09 1.07
(0.08) (0.50) (0.08) (0.49) (0.08) (0.49)
Y68 0.16 1.94 0.17 1.97 0.11 2.09
(0.13) (0.53) (0.12) (0.51) (0.12) (0.54)
Y73 0.50 2.81 0.55 2.89 0.48 2.87
(0.19) (0.70) (0.18) (0.69) (0.18) (0.73)
CHANGE  Y54 0.19 0.14 0.21 0.16 0.14 0.09
(0.18) (0.16) (0.18) (0.15) (0.18) (0.15)
CHANGE  Y63 0.14 0.13 0.17 0.16 0.16 0.14
(0.12) (0.11) (0.12) (0.11) (0.12) (0.11)
CHANGE  Y68 0.30 0.27 0.33 0.30 0.32 0.28
(0.13) (0.12) (0.13) (0.11) (0.14) (0.12)
CHANGE  Y73 0.22 0.24 0.26 0.28 0.21 0.23
(0.20) (0.18) (0.20) (0.19) (0.21) (0.19)
R2 0.37 0.43 0.38 0.45 0.33 0.40
R 2LSDV 0.89 0.90 0.90 0.91 0.89 0.90
No. of industries 71 71 71 71 71 71
No. of industries 16 16 16 16 16 16
with CHANGE ¼ 1
No. of observations 305 305 305 305 305 305
Note: Heteroskedasticity-consistent standard errors in parentheses.
Table 5.12
Regression results for advertising intensity for industries with average ADS > 2%
(fixed-effects estimation)
Dependent variable: ln ADS
ln SSDOM 0.24 0.38 0.24 0.37 0.26 0.39
(0.13) (0.13) (0.13) (0.13) (0.13) (0.13)
ln SSDOM  Y54 — 0.12 — 0.11 — 0.13
(0.07) (0.07) (0.07)
ln SSDOM  Y63 — 0.15 — 0.14 — 0.16
(0.05) (0.06) (0.05)
ln SSDOM  Y68 — 0.19 — 0.19 — 0.20
(0.06) (0.06) (0.06)
ln SSDOM  Y73 — 0.25 — 0.25 — 0.25
(0.07) (0.07) (0.07)
ln K=L 0.07 0.31 0.12 0.32 0.12 0.34
(0.22) (0.21) (0.22) (0.21) (0.21) (0.20)
TVINT 1.34 0.50 — — — —
(0.80) (0.73)
TVTOT — — 0.42 0.31 — —
(0.18) (0.16)
Y54 0.11 0.83 0.13 0.84 0.31 0.83
(0.16) (0.68) (0.13) (0.68) (0.10) (0.67)
Y63 0.03 1.43 0.06 1.34 0.08 1.46
(0.11) (0.56) (0.10) (0.57) (0.10) (0.56)
Y68 0.19 2.09 0.20 2.15 0.12 2.14
(0.18) (0.65) (0.17) (0.65) (0.16) (0.65)
Y73 0.54 3.11 0.59 3.19 0.48 3.14
(0.25) (0.71) (0.25) (0.72) (0.23) (0.71)
CHANGE  Y54 0.25 0.09 0.28 0.13 0.24 0.08
(0.23) (0.15) (0.23) (0.15) (0.23) (0.14)
CHANGE  Y63 0.12 0.16 0.14 0.17 0.12 0.16
(0.15) (0.12) (0.15) (0.13) (0.15) (0.12)
CHANGE  Y68 0.32 0.36 0.35 0.38 0.33 0.36
(0.17) (0.14) (0.17) (0.13) (0.17) (0.14)
CHANGE  Y73 0.20 0.28 0.24 0.31 0.20 0.28
(0.24) (0.20) (0.25) (0.20) (0.25) (0.20)
R2 0.36 0.46 0.37 0.47 0.34 0.45
R 2LSDV 0.86 0.88 0.87 0.88 0.86 0.88
No. of industries 51 51 51 51 51 51
No. of industries 11 11 11 11 11 11
with CHANGE ¼ 1
No. of observations 218 218 218 218 218 218
Note: Heteroskedasticity-consistent standard errors in parentheses.
208 Chapter 5

between market size and advertising intensity observed in the basic


specification is more pronounced in certain years than in others.29
On the whole, there is no evidence of any positive effect of price
competition on advertising intensity, and there is some evidence of
a negative effect. In particular, the coefficient on CHANGE  Y68 is
everywhere negative and statistically significant at the 5% level. On
the other hand, the coefficient on CHANGE  Y73 is also negative,
but it is not statistically significant. The magnitude of the effect of price
competition on advertising is not easy to estimate very precisely. On
the whole, the average effect is a fall in the advertising-sales ratio on
the order of 20–25%. The coefficients on TVINT or TVTOT and the
time dummy for 1973 (in the specification without the interactions of
market size with year dummies) have the expected signs and are
statistically significant. In particular, they confirm our expectations
of a positive effect on ADS of the introduction of TV advertising and
of a negative effect on ADS of the structural changes of the late 1960s.
Note also that the coefficient on CHANGE  Y54, although nega-
tive, is not statistically significant. Recall that this coefficient mea-
sures whether the change in ADS during 1954–1958 was different
between collusive and noncollusive industries. The negative sign of
the coefficient is consistent with the descriptive statistics of table 5.9.
However, there is no statistical significance even at the 20% level,
which strengthens the argument that the observed difference be-
tween the two groups after 1958 is due to the legislation.

Interpreting the Evidence

How should we interpret the results on the effect of the 1956 cartel
law on advertising intensity? As we have seen, the coefficient on
CHANGE  Y68 is negative and statistically significant, while the
coefficient on CHANGE  Y73 is negative but not statistically signi-
ficant. I would argue that the evidence is consistent with the view
that the intensification of price competition following the 1956 Act
probably had a negative effect on advertising intensity across a range
of UK industries during the 1960s. There are three different argu-
ments to support this view.

29. I also tried replacing ln K=L with ln K=N as my setup cost proxy. The results were
very similar to those presented here. They have been omitted, since this variable is not
directly relevant in the advertising regressions anyway.
Price Competition, Advertising, and Market Structure 209

First, recall that the impact of the legislation on competition was


mostly realized during 1958–1968. Second, note that the coefficients
on CHANGE  Y73 are not very much lower in absolute value than
the coefficients on CHANGE  Y68. It is mostly the standard errors
which are higher. This may not be very surprising, since it has been
argued above that there is considerable noise in the data for 1968–
1973. Third, the coefficients on Y68 and especially on Y73 indicate a
significant fall in ADS throughout manufacturing industry during
the late 1960s and early 1970s, which is consistent with the raw sta-
tistics of table 5.10. What table 5.10 also shows is that during 1968–
1973, the average ADS fell by much more in industries not affected
by the 1956 legislation than in industries affected. One reason could
be that for industries in the latter group, advertising intensity had
already fallen considerably during 1958–1968 and could not fall much
farther. For industries in the control group, however, which had not
been much affected during 1958–1968, there was significant scope for
a fall in ADS after 1968, and this fall actually occurred because of
changes in the pattern of distribution or other factors.
Note that this interpretation, although rather speculative, is con-
sistent with the evolution of market structure in the two groups of
industries after 1958, as already discussed above. In particular, it is
consistent with the fact that during 1968–1975, C5 fell slightly, on
average, in advertising-intensive industries not affected by the 1956
Act, although it increased, on average, in advertising-intensive in-
dustries affected by the Act. How could all these different pieces of
evidence fit together?
Suppose that in addition to the effect of the cartel legislation after
1958, which was undoubtedly a major shock, there was an intensi-
fication of price competition across advertising-intensive industries
after the mid-1960s due to the abolition of resale price maintenance
and structural changes in the pattern of distribution. Now consider
the following hypothetical account of events after 1958. During a first
phase, from 1958 to 1968, the 1956 Act caused price competition to
intensify, and hence advertising to fall in previously collusive in-
dustries. However, the fall in advertising was not sufficient, on the
whole, to offset the fall in profits, and so concentration increased in
these industries. During a second phase, from 1968 to 1973, compe-
tition intensified in all advertising-intensive industries, and perhaps
the competitive pressure was again stronger in previously collusive
industries because resale price maintenance had until that time hin-
210 Chapter 5

dered the emergence of fully effective price competition. But these


industries had already experienced a significant fall in advertising as
a result of the cartel legislation, so it was difficult for advertising to
fall much farther. As a result, these industries experienced a further
rise in concentration after 1968. On the other hand, industries that
had not been affected by the cartel legislation responded to the in-
tensification of price competition and the decline in the role of brand
advertising in the late 1960s through a fall in advertising expendi-
ture. This largely offset the pressure on profits and prevented con-
centration from rising in these industries after 1968.
This is not an implausible story, but I should repeat that certain
parts of it are somewhat speculative. In particular, I should empha-
size two main caveats. First, some of the regression coefficients on
which this story relies are not very precisely estimated. Second, the
samples used in the advertising and the concentration regressions
are not identical, and this may affect the comparability of the two
sets of results.
A comparison of the two sets of results, in particular tables 5.5 and
5.6 with tables 5.11 and 5.12, also provides some interesting insights
on the explanatory power of the industry-specific effects in the
regressions presented in this chapter. In all these tables, two R 2 ’s are
reported: the first is derived from transforming the data to obtain
deviations from industry means and applying OLS to the transformed
2
data, while the second (denoted RLSDV ) is obtained from applying
OLS to the untransformed data after including a set of industry
dummies among the regressors. Thus the difference between the two
R 2 ’s is a measure of the explanatory power of the industry effects in
these regressions. Clearly, industry-specific characteristics explain
much of the cross-industry variation in both concentration levels and
advertising intensity. In addition, the concentration regressions have
greater explanatory power than the advertising regressions when the
R 2 is defined to include fixed industry effects and lower explanatory
power when the R 2 does not include industry effects. In other words,
the industry-specific effects have relatively more explanatory power
for concentration than for advertising intensity.
There are at least two possible reasons for this. First, the use of an
imperfect proxy for setup cost in these regressions may imply that
industry effects pick up differences in setup costs across industries,
and these are more directly relevant to concentration than to adver-
tising. Second, market structure is more difficult to change than ad-
Price Competition, Advertising, and Market Structure 211

vertising (a fact reflected in the structure of the game analyzed in


section 5.2), so it is expected to be less responsive than advertising to
relatively small changes in exogenous variables.
Let me now summarize the results so far. I have set out to examine
four empirical questions in this chapter. The econometric evidence
presented in this section has provided answers to three of these
questions. First, the evidence is consistent with proposition 5.1, since
we do not observe the outcomes that are excluded under this prop-
osition. Second, there is considerable evidence of a negative overall
effect of price competition on advertising intensity. Third, there is
strong evidence of a positive overall effect of price competition on
concentration in advertising-intensive industries. The final question,
which concerns propositions 5.2 and 5.3, cannot be answered on the
basis of the econometric results. Support for these propositions is
provided by case studies, and this is the subject of the next section.

5.6 Two Case Studies

This section provides a brief account of the evolution of two different


segments of the UK electrical appliance industry during the 1960s,
focusing on the effect of the intensification of price competition on
advertising and market structure. The overall trend in the evolution
of the electrical appliance industry during the 1960s was an increase
in the intensity of price competition, a fall in advertising intensity,
and a rise in concentration. This evolution is consistent with the eco-
nometric evidence of the previous section. However, even more
interesting than the overall trend was the experience of particular
markets within the industry over specific time periods. This experi-
ence offers several direct tests of the theory developed in this chapter.
The two segments that I will examine are the washing machine
industry and the domestic refrigerator industry. Both industries are
well documented and particularly suitable as case studies for a vari-
ety of reasons. They were relatively concentrated in the late 1950s,
although the market shares of the industry leaders were moderate,
and they were not subject to any restrictions on entry. Moreover,
advertising was a key strategic variable in both industries at the time
the restrictive practices legislation came into effect, as indicated by
advertising-sales ratios around 4% in both cases. Finally, both in-
dustries experienced an increase in the intensity of price competition
during the 1960s, but they were not very much affected by the in-
212 Chapter 5

troduction of TV advertising in the UK. Although the intensification


of price competition was not the only factor that shaped the evolu-
tion of nonprice competition and market structure in the two indus-
tries, this evolution is consistent with the theoretical predictions of
this chapter.
Both industries were subject to restrictive agreements at the time
the 1956 Act came into force (the arrangement on washing machines
was actually part of a wider arrangement covering many types of
electrical appliances). The agreements provided for retail price main-
tenance and common maximum discounts to distributors, and also
contained ancillary restrictions relating to sales promotion and con-
ditions of sale. The refrigerator manufacturers’ agreement also pro-
vided for the exchange of price information between the parties.
There were no explicit restrictions with respect to individual prices.
More important, there were no restrictions on media advertising or
on entry, which is consistent with my theoretical model. Most of the
important washing machine and refrigerator producers were parties
to the agreements. As a result of the 1956 Act, the arrangement on
washing machines was abandoned in 1959 and that on refrigerators
probably was canceled at about the same time. Resale price mainte-
nance continued to be individually imposed until the mid-1960s,
although it was weakening. Eventually this, too, was abandoned
following the introduction of the 1964 Resale Prices Act.
Other exogenous influences on the electrical appliance industry
during the period under study include the discontinuous growth
in demand, the gradual reduction of tariffs and transport costs, and
changes in the pattern of distribution. Demand for all types of elec-
trical appliances was growing, but it was subject to considerable
fluctuations caused to a large extent by changes in macroeconomic
policy (Corley 1966, Hatch 1972). Import penetration was not signif-
icant until the mid-1960s, but it subsequently intensified, for washing
machines and especially for refrigerators. And the breakdown of re-
sale price maintenance in the mid-1960s led to increased competition
among retailers and, together with the growth of large retailers, may
have put additional pressure on manufacturers’ margins or influ-
enced their advertising strategies. Finally, a factor that did not have
much impact on the industry was the introduction of TV advertising:
press advertising remained by far the most important type of adver-
tising for electrical appliances.
Price Competition, Advertising, and Market Structure 213

The Washing Machine Industry

The experience of the washing machine industry from the mid-1950s


to the late 1960s provides a clear contrast between two different com-
petition regimes. The industry was relatively concentrated in 1958,
with the five largest firms accounting for 76% of total sales of UK
firms. The largest producer throughout the 1950s was Hoover, whose
market share had nevertheless declined from about 50% of total in-
dustry sales in the early 1950s to around 35% in 1958. This decline
was to a large extent due to the rapid expansion of A.E.I.-Hotpoint,
achieved largely through heavy media advertising and despite the
fact that the Hotpoint models were among the most expensive in the
market (Mayers 1963, Shaw and Sutton 1976). These two firms had a
combined market share of around 50% in the late 1950s.
The leading firms avoided price competition in the 1950s and pre-
ferred to use advertising to stress product quality, a strategy consis-
tent with the permissive climate of competition policy and also with
the fact that their sales efforts were mainly directed toward upper-
class and middle-class households. Advertising expenditure in the
press and on TV, including production costs of advertisements, in-
creased steadily from 2.5% of total manufacturers’ and importers’
sales revenue in the UK market in 1954 (when TV advertising had
not yet been introduced) to 4.4% in 1958 (with TV advertising ac-
counting for about one-third of total advertising expenditure). Hoover
and A.E.I.-Hotpoint accounted for somewhat less than 40% of total
industry advertising in the UK market.30
The nature of competition in the industry changed dramatically
in the early 1960s. The underlying conditions for this change seem
to have been the introduction of the restrictive practices legislation

30. The advertising data for the whole of this section were taken from the Statistical
Review of Press Advertising, the Statistical Review of Independent TV Advertising, the Sta-
tistical Review of Press and TV Advertising, and the MEAL Monthly Digest of Advertising
Expenditure. The figures were adjusted to take into account press undercoverage, dis-
counts for TV advertising, and production costs of advertisements. Advertising by
accredited dealers is included, together with advertising by manufacturers and
importers, but advertising by the Electricity Council and Electricity Boards (which
have been acting as retailers in the market) for own-brand appliances is not. Data on
sales revenue of UK producers, the five-firm concentration ratio, and firm numbers
were obtained from the Census of Production and Business Monitor, PQ series: Quar-
terly Statistics for various years, and relate to complete machines only. Some data on
market shares were also taken from market research sources. Import and export data
for various years are from the Annual Statement of the Trade of the United Kingdom.
214 Chapter 5

and the emergence of a mass market for washing machines. These


underlying conditions should be distinguished from the event that
triggered the breakout of price competition, namely, the entry of low-
price firms. The most important new entrant was Rolls Razor, a firm
that soon came to represent a serious challenge to established pro-
ducers. Rolls Razor had a three-part strategy (Shaw and Sutton 1976):
offering to sell simple twin-tub machines at prices as much as 25%
lower than incumbent firms; bypassing the traditional distribution
outlets by selling directly to consumers; and engaging in heavy press
advertising both to contact customers and to stress its low prices
(the advertisements contained reply coupons which potential buyers
would use to invite a visit by a salesman). At first, the established
firms’ response was to increase their advertising expenditure and, in
the case of Hoover, also to introduce sophisticated automatic models.
By 1962, however, Rolls Razor had a market share of more than 10%,
and there was entry of more direct-selling, low-price firms, most
notably Duomatic.
Starting in early 1962, the industry leaders cut the prices of their
existing models and also introduced new, cheaper models to com-
pete for the emerging mass market. Advertising expenditure, on the
other hand, soared: in 1963, the industry spent a record 12.6% of
sales revenue in the domestic market on advertising (though if Rolls
Razor and Duomatic are excluded, the industry advertising-sales
ratio was 7–7.5%). Duomatic and Rolls Razor had a combined market
share of 15–18% in 1963, although they accounted for nearly half of
total industry advertising expenditure. On the other hand, Hoover
and A.E.I.-Hotpoint, whose combined market share was 55% in 1963,
accounted for 30% of total advertising expenditure, although they
both had a much higher advertising-sales ratio than in the 1950s.
Note that advertising by the direct-selling firms was of a different
type than that of the established producers, since it was used as part
of a radical distribution policy and as a means of stressing rather
than relaxing price competition. Still, of course, it was a fixed cost
which had to be covered by gross profit.
The fall in the prices of washing machines during the 1960s is
reflected in the official producer price index of the industry. This
increased by 1.5% between 1958 and 1962, then fell by 5% in 1963
and by a further 5.5% between 1963 and 1968. This should be com-
pared with the evolution of the producer price index for all manu-
facturing, which increased by 8% over 1958–1963 and by a further
Price Competition, Advertising, and Market Structure 215

11.5% over 1963–1968. Despite the fall in margins and an increase in


advertising outlays that far exceeded the growth in sales revenue, the
number of UK producers of washing machines employing at least
twenty-five persons increased slightly from twenty-five in 1958 to
twenty-six in 1963, according to the Census of Production. In fact,
there was hardly any merger activity during this period (Hart et al.
1973), and while some exit of smaller firms did occur during the
1961–1962 slump (Hatch 1970), this almost exactly offset the entry of
new firms during the boom of 1959–1960.
Hence the evolution of the industry between 1958 and 1963 seems
to correspond to an outcome which is excluded under proposition
5.1: an increase in advertising outlays without a corresponding fall in
the number of firms following an intensification of price competi-
tion.31 Proposition 5.1 predicts that such an outcome is not sustain-
able in the long run because firms can no longer cover their sunk
costs when faced both with lower margins and with higher adver-
tising expenditure. Note that the growth in demand during 1958–
1963 complicates things somewhat in the present case, but since
advertising expenditure increased by much more than sales revenue,
thus leading to a very large increase in the advertising-sales ratio, the
basic mechanism driving proposition 5.1 still applies.
Events in the washing machine industry after 1963 are consistent
with proposition 5.1. The exit of Rolls Razor in 1964, following a
significant fall in its sales and market share, was only the beginning
of the general restructuring of the industry through exit and merger
(see Hart et al. 1973). The number of UK firms employing at least
twenty-five persons fell from twenty-six in 1963 to eighteen in 1968.
It kept falling after that date, probably because of intensified import
penetration as well as changes in distribution, and was as low as
seven in 1973. The industry advertising-sales ratio also fell after 1964
and was only 2.9% in 1968, with the downward trend continuing
well into the 1970s.32
In summary, the evolution of the washing machine industry in the
early 1960s is interesting for two main reasons. The first reason is

31. The rise of the five-firm sales concentration ratio from 76% in 1958 to 85% in 1963
was mainly due to the 15% market share of Rolls Razor (cf. Hart et al. 1973), and
should not conceal the fact that the structure of the industry remained too fragmented
for firms to be able to cover their sunk costs.
32. Other factors may also have played a role in the restructuring of the industry, such
as the fall in demand between 1963 and 1968, although it must be noted that adver-
tising expenditure fell by much more than sales revenue over that period.
216 Chapter 5

that it helps draw a distinction between long-run effects, which have


been the main focus of analysis in this chapter, and short-run dis-
equilibrium states. The second reason is that it confirms proposition
5.1 in a very direct way, namely, by documenting the mechanism
that breaks an unsustainable industry configuration: when firms
cannot cover their sunk costs, there must be exit or a reduction in
endogenous sunk costs or both. More specifically, the short-run dis-
equilibrium appears in this case to correspond to that outcome which
is excluded under proposition 5.1, namely, an increase in advertising
without a corresponding fall in the number of firms following an
intensification of price competition. The evolution of the washing
machine industry shows that this outcome can occur in the short run,
but it must eventually break down, because it is not sustainable in
the longer term.

The Domestic Refrigerator Industry

The evolution of the domestic refrigerator industry since the mid-


1950s provides further support for the theoretical results of this
chapter, in particular propositions 5.2 and 5.3. Domestic refrigerators
can be mechanically operated or heat-operated. The former, which
accounted for about 65% of total sales in 1958 and about 75% in 1968,
are all electric. Of the latter, which were in decline throughout the
period under study and accounted for less than 10% of total sales in
1973, most are electric but some are gas models. The relevant market
is the market for all types of refrigerators, because the gas models
have always been coproduced with their electric counterparts, and
the restrictive agreement of the refrigerator manufacturers covered
all types.
The industry was relatively concentrated in 1958, with only nine or
ten UK firms employing at least twenty-five persons, although none
of these had a market share higher than 20%. A change of competi-
tion regime seems to have occurred soon after the 1956 Act came into
effect. The emergence of price competition was triggered by the
slump of 1961–1962; this left many firms in the industry with con-
siderable excess capacity, some of which had been the result of ex-
pansion by established producers or entry of new firms during the
boom of 1958–1960. Net producer prices, which had been relatively
stable throughout the 1950s, fell by about 20% between 1960 and
1964 (Hatch 1970, Mayers 1963). This is also reflected in the official
Price Competition, Advertising, and Market Structure 217

producer price index of the industry, which fell by more than 25%
between 1958 and 1963, while the producer price index for all man-
ufacturing increased by 8% over the same period.
Changes in distribution in the mid- and late 1960s, including the
growth of retailers’ own brands (Baden-Fuller 1980) may have put
further pressure on profit margins. An additional major influence on
competitive conditions in the industry since the mid-1960s was the
growth of imports. While in the 1950s these were not significant, in
1968 they accounted for more than a quarter of total manufacturers’
and importers’ sales revenue in the UK market. The producer price
index of the industry remained unchanged between 1963 and 1968,
at a time when the producer price index for all manufacturing
increased by about 11.5%.
Nonprice competition was an important element of manufacturers’
strategies in the 1950s, but its significance diminished over the fol-
lowing decade. Advertising expenditure in the press and on TV was
3.7% of total manufacturers’ and importers’ sales revenue in the do-
mestic market in 1958 and remained high during the early 1960s. The
most important advertisers throughout this period were Electrolux,
Frigidaire, and, until its closure in 1964, Pressed Steel. These three firms
were consistently among the four largest producers of refrigerators in
the 1950s. Note that part of the reason for the high advertising-sales
ratio of the industry in the early 1960s was a 1962 deal between Rolls
Razor and Pressed Steel, whereby Rolls Razor would market Pressed
Steel refrigerators. The venture accounted for more than a quarter of
the industry’s advertising expenditure in 1963, but for only about 7%
of sales, and Pressed Steel eventually ceased production. Thus it
seems reasonable to regard the high advertising of the early 1960s as
a short-run disequilibrium response to the new competitive regime
in the industry.
This interpretation is consistent with the fact that the industry ad-
vertising-sales ratio collapsed after the exit of Pressed Steel in 1964,
reaching 1.1% in 1968 and even lower levels in the 1970s.33 One rea-
son for this large fall in advertising intensity was the intensification
of price competition. Another reason must have been the negative
effect of the changes in distribution and the growth of retailers’ own
brands on the effectiveness of brand advertising by manufacturers.

33. Advertising by domestic producers never recovered. A small temporary increase


in industry advertising intensity in the early 1970s was largely due to advertising of
certain imported brands.
218 Chapter 5

How was the evolution of market structure in the refrigerator in-


dustry affected by the described changes in the competition regime
and the role of advertising? Between 1958 and 1968, price competi-
tion intensified and advertising intensity fell dramatically. According
to proposition 5.2, these are circumstances where a fall in concentra-
tion (a rise in firm numbers) is possible. In particular, if we were to
observe a decline or no change in concentration in the refrigerator
industry between 1958 and 1968, then proposition 5.2 would be
confirmed. This is because the proposition predicts that whenever
concentration does not rise following an intensification of price
competition, then it must be the case that advertising has fallen sub-
stantially, everything else being equal. Note that if, on the other hand,
we were to observe a rise in concentration in the refrigerator industry
between 1958 and 1968, proposition 5.2 would not be rejected; one
could always claim that the fall in advertising was not large enough.
However, the empirical relevance of proposition 5.2, and more gen-
erally of the idea that a negative competition effect on concentration
is possible in advertising-intensive industries, would be seriously
questioned, since it would then be difficult to see under what plau-
sible industry conditions we may get an outcome different from the
standard positive competition effect on concentration.
There is also a slightly different way to think of this particular
experiment, which makes use of proposition 5.3 as well. The refrig-
erator industry was one of a handful of industries with an initial
advertising-sales ratio higher than 3% that experienced a significant
increase in the intensity of price competition in the 1960s. A high
advertising-sales ratio implies that there is scope for a significant
reduction in advertising, and thus for a negative effect of price
competition on concentration. Of all the industries in this class, the
refrigerator industry experienced one of the largest falls in the
advertising-sales ratio during the 1960s. If the hypothesis of a poten-
tially negative competition effect on concentration in industries with
high advertising effectiveness is to have any empirical relevance,
one should see this effect in this particular industry.
In fact, one does see a negative competition effect on concentra-
tion, and a positive effect on the number of firms, in the refrigerator
industry between 1958 and 1968. Changes in market size and other
factors complicate somewhat the interpretation of the evidence in
terms of our theoretical predictions, but the basic mechanisms at
Price Competition, Advertising, and Market Structure 219

work should still be reflected in the evidence.34 Thus the number of


domestic firms did not change significantly over 1958–1968. Accord-
ing to the Census of Production, the number of UK manufacturers of
mechanically operated domestic refrigerators employing at least
twenty-five persons was nine in 1958, and this number was the same
in 1963 (despite the considerable entry and exit of firms over 1959–
1963; see Hatch 1970) and in 1968. These figures do not include for-
eign firms selling in the UK market. Given that imports were far
higher in 1968 than in 1958, the number of foreign firms with sales
in the UK must have been higher in 1968 than in 1958, and hence the
same must also have been the case for the total number of firms
(excluding those with fewer than twenty-five employees) with sales
in the UK. Note that the years 1958, 1963, and 1968 are all broadly
comparable with respect to demand conditions.
Moreover, evidence from market research suggests that the four-
firm concentration ratio in the UK market for domestic electrical
refrigerators as a whole (including imports) declined from about 60%
in the late 1950s to 45% in 1967 before rising to 55% in 1969 (see
Baden-Fuller 1980). Note that much of the rise in concentration be-
tween 1967 and 1969 was due to the 1967–1968 A.E.I.-G.E.C.-English
Electric mergers, which were not carried out with the specific objec-
tive of merging the electrical appliance divisions of these firms.35
One reason for the fall in concentration during the 1960s was a
shifting of market share away from the two leading domestic firms,
something not really captured by the model of this chapter (which
assumes symmetry). However, another reason was the growth of
imports, so the fall in concentration is also linked to the evolution of
firm numbers and thus can be attributed partly to the forces identi-
fied in the theoretical model.

34. The effect of market growth is to create a ‘‘bias’’ toward a rise in firm numbers and
a fall in concentration to the extent that the exogenous component of sunk costs is
significant. However, this may not be true in a high-advertising industry, since market
growth will then induce an escalation of advertising expenditure, all else being equal
(see Sutton, 1991). Moreover, other factors may influence concentration upward and
firm numbers downward, as is likely to have happened in the present case—see the
discussion in chapter 4 on the various factors that contributed to the overall rise in
concentration in the UK during the 1960s.
35. These mergers had a much smaller impact on concentration in the washing
machine industry than in the refrigerator industry because of the low market share of
G.E.C. in washing machines and a 1965 agreement whereby A.E.I.-Hotpoint took over
English Electric’s washing machine division and English Electric took over A.E.I.-
Hotpoint’s stove division (see Hart et al. 1973).
220 Chapter 5

It should be pointed out that the concentration data are not strictly
comparable with the data on firm numbers described above. The for-
mer refer to sales of all electrical refrigerators in the UK market; the
latter refer to sales of mechanically operated refrigerators by UK pro-
ducers employing at least twenty-five persons. Nevertheless, both
pieces of evidence point in the same direction. Moreover, the stabil-
ity, or even increase, of firm numbers and the apparent fall in con-
centration in the refrigerator industry between 1958 and 1968 should
be seen in the context of the overall substantial fall in firm numbers
and the increase in concentration in British manufacturing as a whole
during the 1960s. As pointed out in chapter 4, Hart and Clarke (1980)
report a rise of eight percentage points in the average five-firm sales
concentration ratio among UK producers at the four-digit industry
level between 1958 and 1968. My own calculations suggest that most
of this can be attributed to a fall in firm numbers, since the number of
UK firms employing at least twenty-five persons fell during the same
period by about 25–30% in the average industry (see chapter 7 of this
book). Compared to this benchmark, the refrigerator industry expe-
rienced a relative rise in firm numbers and a relative fall in concen-
tration during 1958–1968, even if foreign firms are not taken into
account.
Interestingly, this is not the end of the story for the refrigerator
industry. Competitive pressure continued to intensify after 1968, as
shown by the fact that the producer price index of the industry
increased by only 9.5% over 1968–1973, while the producer price in-
dex for all manufacturing increased by 37% over the same period.
Several factors may have caused price competition to intensify fur-
ther, and one of them was certainly import penetration: imports
increased from 25% of total sales in the UK in 1968 to almost 50% in
1973. Now advertising was already low in 1968, so it could not fall
much farther. Proposition 5.2 would then predict that during 1968–
1973 the number of firms should have fallen, everything else being
equal, since the pressure on margins could no longer be offset by any
significant reduction in sunk costs. This is indeed what happened:
according to Business Monitors, there were only six UK producers of
all types of domestic refrigerators still in operation in 1972–1973.
In summary, the experience of the refrigerator industry during
1958–1968 shows that a negative competition effect on concentration
is not just a theoretical oddity. This was a high-advertising industry
that experienced a significant increase in the intensity of price com-
Price Competition, Advertising, and Market Structure 221

petition and a large fall in advertising expenditure during the 1960s.


These are the conditions that can lead to a fall in concentration, and
in this case they did. Furthermore, both phases in the evolution of
the refrigerator industry seem to be consistent with proposition 5.2.
The experience of the industry during 1958–1968 suggests that an
increase in the intensity of price competition must lead to a substantial
fall in advertising if the number of firms does not fall (and concentra-
tion does not rise), as predicted by the proposition. The experience of
the industry during 1968–1973, on the other hand, suggests that an
intensification of price competition must cause the number of firms
to fall (and concentration to rise) if advertising falls little or not at all,
which is again consistent with proposition 5.2.

5.7 Concluding Remarks

In advertising-intensive industries, tougher price competition will


affect both the incentive of firms to spend on advertising and market
structure. The theoretical predictions on the effects of competition in
this class of industries are weaker than the corresponding results for
exogenous sunk cost industries. The theory is in general silent as to
whether advertising expenditure will rise or fall when price compe-
tition intensifies. Moreover, if advertising rises, then concentration
must also rise. But if advertising falls, then the decrease in the sunk
costs that firms must cover at equilibrium may or may not offset the
fall in gross profits due to increased price competition, and so con-
centration may decrease or increase.
In the first part of this chapter a theoretical framework was intro-
duced to deal with these issues. The focus was on deriving predictions
on the joint effect of price competition on advertising and market
structure. Two general results were derived. The first of these places
a bound on the space of outcomes by excluding certain outcomes as
not admissible within the theory. In particular, a fall in concentra-
tion together with an increase in advertising expenditure, follow-
ing an increase in the intensity of price competition, cannot occur.
The second result, which further restricts the space of outcomes, is a
prediction on the competition-concentration relationship which is
conditional on the behavior of other observable variables. The key
implication of this result is that more price competition must lead to
a significant fall in advertising intensity if concentration does not
rise, or, conversely, it must cause concentration to rise if advertising
222 Chapter 5

falls little or not at all. Taken together, the theoretical results indicate
that a positive effect of price competition on concentration is the
most likely outcome in advertising-intensive industries, since the re-
verse can occur only in the event of a significant fall in advertising
expenditure.
In the second part of this chapter, the theory was tested against
econometric and case-study evidence on the evolution of advertising-
intensive manufacturing industries following the introduction of
cartel legislation in Britain in the late 1950s. The econometric results
indicate that the intensification of price competition following the
introduction of the legislation caused, on the whole, a rise in con-
centration of about six percentage points in advertising-intensive
industries, and probably also a fall in advertising intensity on the
order of 20–25%. The case-study evidence confirms that a negative
effect of price competition on concentration cannot be ruled out in
high-advertising industries. However, a necessary condition for this
to occur is that tougher price competition leads to a significant fall in
advertising intensity. On the other hand, if advertising falls little or
not at all when price competition intensifies, then concentration will
rise, which is consistent with the theoretical predictions as well as
with the econometric results.
6 Price Competition,
Innovation, and Market
Structure in R&D-
Intensive Industries

6.1 Introduction

The relationship between market power and innovative activity has


been a much-debated issue ever since Schumpeter’s pioneering work.
On the one hand, large firms in concentrated markets are often seen
as the main engines of technological progress, for reasons that relate
to the optimal scale for R&D and innovation, appropriability con-
ditions, and the presence of financial constraints. For instance, one
argument along these lines is that in the presence of capital market
imperfections, firms with market power can more easily finance R&D
from own profits and are therefore more likely to undertake R&D
projects, which are inherently risky and often involve significant sunk
costs. Another argument is that firms with market power can more
easily appropriate the returns from innovation, and therefore have
better incentives to innovate.
On the other hand, it is often argued that the lack of competitive
pressure may lead to inertia and managerial slack, and hence to a
reduced level of innovative activity. In its simplest version, this view
essentially implies that managers and workers of firms with market
power can capture some of the monopoly rents in the form of slack
or lack of effort. More recent versions of this idea have focused on
the fact that competition may improve the ability of firms’ owners to
provide appropriate incentive schemes to managers.
The view of competition as a stimulus to innovation is probably
the majority position among policy makers today. This represents a
significant shift in opinion from the days when promoting national
champions was regarded by many as the best way to encourage in-
novation. But although this latter position seems to be in decline, the
224 Chapter 6

theoretical and empirical support for the view that competition is


generally beneficial for innovation is far from clear. In fact, despite
the very extensive literature on the subject, the issue of the links
between market power and innovation is still not settled.

A Long-Standing Debate

The theoretical literature on competition and innovation consists of


several strands. One line of research has focused on the interaction
between short-run and long-run decisions in oligopoly. Although
some of these studies have found that less short-run competition
may result in, or be associated with, more investment in the long-run
variable (such as R&D expenditure), others have shown this result
to depend on particular functional forms or the specification of the
collusive technology (see Yarrow 1985, Fershtman and Gandal 1994,
Ziss 1994, Fershtman and Muller 1986, Davidson and Deneckere
1990, Fershtman and Pakes 2000).1 Another strand of the theoretical
literature has analyzed endogenous growth models of innovation.
These studies have also produced mixed overall results, although
they have succeeded in identifying specific conditions that may de-
termine whether competition is associated with more or less innova-
tion. Thus, for instance, a negative competition effect on innovation
has been obtained in models with profit-maximizing firms and leap-
frogging technological progress, while a positive effect can occur if
firms are non-profit-maximizing or technological progress is step-by-
step (see Aghion and Howitt 1997, 1998; Aghion et al. 1999). Note
that strong predictions regarding the effect of competition on innova-
tion are generally more difficult to derive in models with asymmetric
firms, since the response of each firm to intensified competition may
then depend on the relative efficiency levels (see Boone 2000). Finally,

1. A common assumption in this context is that firms cannot collude in the long-run
variable, whatever their short-run conduct. This is often justified in theoretical studies
by a reference to the fact that long detection and retaliation lags hinder the stability of
collusion. Since long-term decisions take time to implement, the reaction to rivals’ be-
havior is relatively slow (i.e., there are relatively long retaliation lags); hence collusion
in long-run variables will be relatively difficult to achieve. In the case of R&D, an ad-
ditional argument is that deviations from agreed levels of R&D expenditure are diffi-
cult to observe, at least as long as R&D cooperation does not take the form of a joint
venture; in other words, detection is also problematic, and this further hinders collu-
sion on R&D.
Price Competition, Innovation, and Market Structure 225

the theoretical literature on the effects of competition on managerial


incentives has provided only weak support for the view that com-
petition improves managerial performance (see Nickell 1995 for a
survey).
On the empirical side, a number of surveys (e.g., Cohen 1995,
Symeonidis 1996) suggest that there is no strong general relationship
between market power and innovation, and that industry charac-
teristics such as appropriability conditions, demand, and especially
technological opportunity explain much more of the cross-industry
variation in R&D intensity and innovation than market power or
market structure. It is therefore all the more interesting that the
results from a number of UK studies of innovation seem to be less
ambiguous. In particular, both Geroski (1990) and Blundell et al.
(1995) have found evidence of a positive link between competition
and innovation in British industry.
One of the main methodological problems in the empirical litera-
ture on the determinants of innovation has been the relative neglect,
until recently, of the endogeneity of market structure. Thus an im-
plicit assumption in many empirical studies has been that market
power is greater in concentrated markets. This assumption implies
that to analyze the effect of market power on innovation, one can
simply examine the effect of concentration or firm numbers on inno-
vation. This approach, however, is not justified in a context where
market structure is seen as endogenous. Within the theoretical
framework of this book—which centers on the idea that more intense
price competition will generally lead to higher concentration—an
approach that considers high concentration as a proxy for high mar-
ket power is simply wrong. Moreover, the empirical evidence pre-
sented in chapters 4 and 5 suggests that this criticism should be
taken seriously.
More specifically, there are two main implications of the theoreti-
cal and empirical analysis of previous chapters of this book for the
study of the links between market power and innovation. First,
market structure cannot be taken as a proxy for market power in
cross-industry studies. Second, both innovation and market structure
should be seen as jointly determined endogenous variables, a view
which is consistent with much recent work on technological change
and the evolution of market structure in R&D-intensive industries
226 Chapter 6

(see Utterbach and Suarez 1993, Klepper 1996, Klepper and Simons
1997, Sutton 1996, 1998, among others).2
There have been numerous attempts to address one or the other
of these issues in econometric work. Thus some authors have ex-
amined the determinants of innovation using measures of compe-
titive pressure or market power other than market structure, such
as import penetration; some of this work will be reviewed below.
Others have tried to address the endogeneity issue by using instru-
mental variable techniques or by estimating simultaneous equation
systems, with rather mixed results (see, e.g., Farber 1981, Lunn 1986,
Levin and Reiss 1988).3 However, none of these studies has explicitly
modeled innovative output and concentration as jointly determined
by some exogenous measure of competition (as well as by other
variables). This is the approach that will be taken here.
Before describing my methodology in more detail, I will briefly
discuss a number of previous empirical studies of the determinants
of R&D or innovation that have used measures of market power
other than market structure. Taken as a whole, these studies provide
little evidence of a negative effect of competition on innovation, and
only weak evidence of a positive effect.
This is certainly the conclusion from a comparison of two studies
of the effects of import competition on R&D or innovation. The first
is a detailed study of R&D-spending reactions of US firms to in-
creased high-technology import competition over the period 1971–
1987, carried out by Scherer and associates (see Scherer 1992, Scherer
and Huh 1992). The authors found that the response varied consid-
erably across firms, and that most of this variation was unsystematic.

2. This recent literature consists of two main strands. One line of research has focused
on developing stylized theories of the life cycle of R&D-intensive industries and test-
ing them against evidence from industry case studies. See Klepper and Simons (1997)
for a comparison of different theories. A second approach has questioned the general-
ity of any particular evolutionary path, and has focused instead on a small number of
mechanisms that are likely to work across a broad range of industries. This is the
approach of Sutton (1996, 1998).
3. Innovation may affect market structure directly or indirectly, and in various and
contradictory ways. For instance, it may increase or decrease the minimum efficient
scale of production, and it may facilitate or hinder the entry of new firms. So it is per-
haps not surprising that the econometric literature on this issue has been inconclusive.
For instance, while Farber (1981) and Levin and Reiss (1988) found a positive effect of
R&D on concentration, Mukhopadhaya (1985) found a negative effect of R&D on con-
centration change, and Geroski and Pomroy (1990) found a negative effect of innova-
tion counts on concentration change.
Price Competition, Innovation, and Market Structure 227

If anything, rising import competition had, on average, a negative


but statistically insignificant effect on R&D intensity in the short run.
In the longer term, no overall effect either way could be identified,
partly because of data limitations. On the other hand, Bertschek (1995)
found a positive and statistically significant association between the
import penetration ratio and the introduction of innovations in a
panel of German firms over the period 1984–1988. This study focused
on short-run reactions only, and left open the question of the long-run
effect of changes in the competitive environment on innovation. More
generally, a difficulty with studies of the effects of import penetration
is the potential endogeneity of imports and the possible presence of
reverse causality between import penetration and innovative activity.
A more long-run view was taken in Geroski’s (1990) study of the
determinants of innovations used (as opposed to innovations pro-
duced) at the three-digit industry level in the UK during the 1970s.
Geroski split the 1970s into two five-year periods, and exploited the
resulting panel structure to control for technological opportunity and
other time-invariant industry-specific characteristics. He included six
different measures of market power among his independent vari-
ables: the extent of market penetration by entrants, the number of
firms with fewer than ninety-nine employees as a fraction of the total
number of firms, the market share of exiting firms at the year of exit,
the five-firm concentration ratio, the change in the concentration ratio,
and import penetration. The coefficient on import penetration was
negative but not statistically significant. The coefficients on the other
five variables, however, suggested that more competitive conditions
favored the use of innovations by firms, although only in the case of
the concentration ratio and the change in the concentration ratio was
the effect statistically significant.
While Geroski found evidence of a positive effect of competition
on the use of innovations across British manufacturing industries in
the 1970s, a more recent study by Broadberry and Crafts (2000) found
no evidence of any link between the two variables in the 1950s. The
measure of competition used by Broadberry and Crafts was based on
the incidence of collusive pricing in the 1950s, and their results were
robust to different ways of classifying industries as collusive or com-
petitive. A potential limitation of their analysis derives from the use of
cross-section data. As will be pointed out in section 6.3, the available
innovations data are subject to measurement errors, which may make
their use in cross-section regressions—but not in panel regressions—
228 Chapter 6

somewhat problematic. Furthermore, it is generally very difficult to


unravel the two-way link between competition and innovation using
cross-section analysis. These difficulties can, however, be avoided by
examining the evolution of the two variables over time—in particu-
lar, by examining the evolution of innovative activity across industries
following the enactment of the 1956 restrictive practices legislation.

The Present Study

This chapter focuses on industries where R&D and technological


innovations are important for competition. The theoretical frame-
work used to analyze these industries has much in common with
that developed in chapter 5 for analyzing advertising-intensive in-
dustries. Equilibrium in R&D-intensive industries is seen here as
involving the joint determination of market structure, R&D expen-
diture, and innovative output. As elsewhere in the book, the inten-
sity of price competition can be thought of as an inverse measure of
the degree of collusion, and is taken as exogenous.
I will begin by adapting the general theoretical framework of
chapter 5 to the analysis of R&D-intensive industries. The model is
sufficiently general to allow for a variety of possible outcomes re-
garding the effect of price competition on innovation. This approach
is motivated by the idea that strong results on this question can be
obtained only by imposing a considerable amount of structure on
theoretical models. Such results are often of limited generality or
depend on features of the model that are difficult to observe or mea-
sure across industries. My approach is also motivated by the incon-
clusiveness of much of the existing literature on the links between
market power and innovation. Instead, I will focus on weaker pre-
dictions that can be derived from a more general model and relate
to the joint behavior of innovation and market structure following
an intensification of price competition in industries with significant
R&D expenditures.
In the second part of the chapter, I will examine the econometric
evidence on the effect of price competition on concentration and the
production (rather than the use) of innovations in the UK following
the introduction of the 1956 Restrictive Trade Practices Act. Although
the effect of the 1956 Act on innovation has often been discussed (see,
e.g., Wilson 1962, Swann et al. 1973, 1974), this chapter provides the
first systematic analysis of this issue.
Price Competition, Innovation, and Market Structure 229

My empirical methodology differs from that of previous empirical


studies of the links between market power and innovation. In partic-
ular, I model market structure and innovative output as endogenous
variables in reduced-form equations derived from a game-theoretic
model. I make no attempt to analyze the interaction between inno-
vation and market structure using a simultaneous-equation system,
since the available data are simply not sufficient for obtaining proper
identification of the equations in such a system. I use the panel struc-
ture of the data to control for industry-specific effects as well as for
key determinants of innovation, such as technological opportunity,
which are likely to be correlated with measures of competition but
are relatively stable over time. Moreover, I explicitly focus on long-
run effects, although I also compare these to short-run effects. Finally,
and perhaps most important, I bypass the need to measure or proxy
the intensity of price competition, since I use information on a major
exogenous institutional change that significantly affected the com-
petitive environment facing UK firms in several industries.

6.2 Theoretical Framework

Consider an R&D-intensive industry where each firm produces one


or more varieties of a differentiated product. A three-stage game can
be used to model competition in such an industry. At stage 1 firms
decide whether or not to enter the industry at an exogenously given
cost of entry f . At stage 2 each firm i chooses to incur a cost Ri , which
represents expenditure on process or product R&D. The results of
R&D are measured by an index of ‘‘innovative output’’ Ii , which
is also realized at stage 2 of the game. A rise in Ii enhances product
quality (and thereby increases the consumers’ willingness to pay) or
reduces the marginal cost of production; either way, it affects the
firms’ objective functions in the final-stage subgame. Finally, at stage
3 firms simultaneously set prices or quantities. I will be looking, in
what follows, for a symmetric subgame perfect equilibrium in pure
strategies in this game.4

4. The structure of the model analyzed in this section is similar to the structure of the
linear demand model of section 5.3. The competition measure t is the equivalent of
the inverse of the parameter l of the linear demand model. The innovation index I is
the equivalent of the quality index u, and the basic technological relationship between
Ii and Ri is similar to that between ui and Fi in the linear demand example. Finally,
assumptions R1–R3, made below, are all satisfied in the linear demand example. Of
course, the present model is much more general.
230 Chapter 6

The structure of this game reflects the fact that (1) decisions on
R&D expenditure involve significant sunk costs and (2) product or
process characteristics cannot be changed as easily or as quickly as
price choices, although they are still more flexible than entry deci-
sions. However, I abstract from issues of uncertainty regarding the
outcome of R&D projects, as do, in fact, many models within the
general class of ‘‘nontournament’’ models of R&D rivalry, of which
the present model is an example.5 The key underlying assumption of
these models, as opposed to the literature on patent races, is that
each firm can achieve product or process improvements through its
R&D expenditure without preventing other firms from obtaining
equivalent benefits from their own R&D.

The Third-Stage Subgame

The equilibrium of the third-stage subgame can be represented by a


vector of gross profits pi ðS; N; h; t; I1 ; . . . ; Ii ; . . . ; IN ; k1 ; . . . ; ki ; . . . ; kN Þ,
where S is market size, N is the number of firms that have entered at
stage 1, h is a measure of the degree of horizontal product differen-
tiation, ki is a vector of parameters specific to firm i and independent
of N and the Ii ’s, and t is a measure of the intensity of price compe-
tition.6 The interpretation of t is the same as in previous chapters: for
any given N, gross profit pi will depend on the firms’ pricing behavior,
which in turn will depend partly on exogenous institutional factors.
For instance, t can be thought of as an inverse measure of the degree
of collusion.
In the benchmark case of symmetric firms, the equilibrium gross
profit of firm i can be written as pi ðS; N; h; t; Ii ; ðIi ÞÞ, where ðIi Þ denotes

5. Previous studies using nontournament models of R&D competition in which mar-


ket structure and the level of R&D are jointly endogenous variables include Dasgupta
and Stiglitz (1980), Motta (1992), and Sutton (1996, 1997b, 1998).
6. In R&D-intensive industries the degree of horizontal product differentiation depends
not only on demand characteristics, such as the extent of diversification of users’ pref-
erences and the level of transport costs, but also on technological factors—namely, the
availability of alternative research paths leading to different varieties of the product or
associated with different submarkets within the industry, as well as the extent of scope
economies in R&D across the various research paths (see Sutton 1998). For our present
purposes, it will be sufficient to subsume all these influences—which can be regarded
as exogenous—within the concept of horizontal product differentiation. This is not
inconsistent with Sutton’s emphasis on the trade-off between spending on R&D to
enhance the quality of existing products and spending to develop new products, and
the implications of this trade-off for market structure in technologically progressive
industries. In the present model, the incentive to provide variety increases (and the
equilibrium level of concentration should normally fall) as h rises.
Price Competition, Innovation, and Market Structure 231

the ðN  1Þ-tuple ðI1 ; I2 ; . . . ; Ii1 ; Iiþ1 ; . . . ; IN Þ. Alternatively, one can


write gross profit as pi ½S; CðN; Ii ; ðIi ÞÞ; h; t; Ii ; ðIi Þ, where C is any
concentration measure whose value increases in 1=N for given Ii ’s
(and depends only on N when Ii ¼ Ij , Ei; j, j 0 i), such as the concen-
tration ratio. I will assume that pi is everywhere increasing in S and h,
and decreasing in t and N. Moreover, I will make two assumptions
about the effect on pi of changes in own and rival innovative output:
Assumption R1. pi is increasing and strictly concave in Ii .

Assumption R2. pi is nonincreasing in Ij , Ej 0 i.


The first of these assumptions says that a firm’s gross profit is in-
creasing and strictly concave in its own innovative output, while the
second says that a firm’s gross profit is nonincreasing in any given
rival’s innovative output.

The Second-Stage Subgame

At stage 2 each firm chooses Ri to maximize its net profit pi  Ri  f ,


given the choices of the other firms and the number of firms that
have entered at stage 1. Now assume, for simplicity, that the choice
of a level of expenditure Ri by firm i entirely determines the inno-
vative output of that firm, given the fundamental (and exogenous)
technological characteristics of the industry. That is, let innova-
tive output Ii be determined by the innovation production function
Ii ¼ Iðr; Ri Þ, where r is a measure of technological opportunity (hence
Ii is strictly increasing in r for any given value of Ri > 0). Note that
this function implies that Ii is independent of Rj , Ej 0 i (i.e., that there
are no R&D spillovers). This assumption is made for simplicity and
could be relaxed without changing the key results of the model, pro-
vided that spillovers are not very large, so that the gross profit pi
remains nonincreasing in Rj , Ej 0 i.7

7. To see this, note that the R&D of firm j has two opposing effects on the gross profit
of firm i. On the one hand, an increase in the R&D of firm j raises its innovation index,
and thus increases its market share and gross profit at the expense of firm i’s market
share and gross profit. This negative R&D externality is always present in the model.
On the other hand, when there are positive R&D externalities (spillovers), there is also
another effect: an increase in the R&D of firm j raises the innovation index of firm i, and
thus causes the gross profit of that firm to increase at the expense of firm j’s own gross
profit. The results of the model are not altered as long as this second effect does not
dominate the first (i.e., as long as spillovers are not too large). It is interesting in this
respect that Geroski (1994) found no evidence of significant R&D spillovers in British
industry during the period that I analyze in this book.
232 Chapter 6

Another implicit assumption, which is probably more difficult to


relax without modifying the results, is that the ‘‘efficiency’’ of firms in
carrying out R&D projects does not depend on the intensity of price
competition. In other words, for any given level of Ri , Ii does not
depend on t. Finally, the function Ii ¼ Iðr; Ri Þ also abstracts from any
stochastic influences on innovative output. This may seem like an
oversimplification, but it merely reflects the fact that uncertainty
does not play an important role for the issues examined here.
A key implication of the above innovation production function is
that R and I are essentially interchangeable choice variables. In other
words, it makes no difference whether the equilibrium of the second-
stage subgame is defined in terms of a set of conditions for the opti-
mal choice of the Ri ’s or in terms of a set of conditions for the optimal
choice of the Ii ’s. Finally,

Assumption R3. Ii is increasing and weakly concave in Ri , with


dRi =dIi < dpi =dIi at Ri ¼ 0, and there exists a level of Ri such that
dpi =dIi ¼ dRi =dIi for any given N and set of Rj ’s.

The weak concavity of the innovation production function is actu-


ally a stronger condition than what is required, but it is consistent
with the empirical evidence, which provides strong support for the
presence, on average, of constant or diminishing returns to scale in
the production of innovations (see, e.g., Kamien and Schwartz 1982,
Bound et al. 1984, Hausman et al. 1984, Crepon and Duguet 1997).
The results of this section would not change if the innovation func-
tion were convex, provided that gross profit pi remained strictly con-
cave in Ri .8 The second part of assumption R3 ensures that there exists
a level of Ri > 0 such that one more unit of R&D costs as much as the
gross profit it creates (i.e., dpi =dRi ¼ 1, Ei). This is a necessary condi-
tion for the existence of an interior solution to the firm’s problem of
choosing the optimal innovation index (or the optimal level of R&D
expenditure).
The first-order condition for the optimal choice of Ii by firm i is
dpi =dIi ¼ dRi =dIi . This says that for any given N and set of Ij ’s, each

8. The requirement that dRi =dIi < dpi =dIi at Ri ¼ 0 ensures that firms will choose to
carry out some R&D at equilibrium. The assumption of a continuous innovation pro-
duction function is, of course, a simplification. A more general formulation would
allow for the presence of indivisibilities in R&D projects, especially since in some
industries the absolute size of the typical R&D project may be large. However, the
existence of indivisibilities does not significantly alter the basic analysis of the incen-
tives to conduct R&D.
Price Competition, Innovation, and Market Structure 233

firm spends on R&D up to the point where the cost of an extra unit of
R&D is equal to the profit this creates in the third-stage subgame
through the corresponding rise in innovative output. At any sym-
metric equilibrium, we have Ii ¼ I (and hence Ri ¼ R), Ei, so the first-
order condition for firm i can be written as

d d
pi ðS; N; h; t; Ii ; ðIi ÞÞ ¼ Ri ðr; Ii Þ at Ii ¼ Ii ¼ I: ð6:1Þ
dIi dIi

Equation (6.1) defines the level of R&D expenditure incurred by


each firm, and the level of innovative output, as a function of the
number of firms that have entered at stage 1. Note that at stage 2 of
the game the Ii ’s and Ri ’s are the Nash equilibrium outcomes of a
noncooperative game, whatever the value of t at the final stage. This
assumption is standard in many theoretical models of semicollusion,
and it is also consistent with the empirical evidence presented in
chapter 2 of this book, according to which collusion with respect to
R&D was rare in British manufacturing in the 1950s, even in indus-
tries where firms colluded on price.9

The Symmetric Equilibrium

At stage 1 firms enter as long as they can make a nonnegative net


profit, anticipating equation (6.1). The free-entry condition, which—
as in previous chapters—is assumed to hold whatever the value of t
may be at the final stage of the game, is given by

pi ¼ pi ðS; N; h; r; t; Ri ; ðRi ÞÞjRi ¼Ri ¼R ¼ R þ f ; Ei; ð6:2Þ


where, for simplicity, N is treated as a continuous variable. Note that
the gross profit has been written as a function of the Ri ’s and r (rather
than the Ii ’s) in equation (6.2). This way of expressing the profit

9. It is, of course, true that firms may find it easier to collude on R&D when R&D co-
operation takes the form of a joint venture. However, as pointed out in chapter 2, there
was hardly any R&D cooperation of that form in British industries that practiced price-
fixing in the 1950s. More generally, allowing for collusion on R&D in the present
model would probably increase the likelihood that a breakdown of collusion (affecting
all choice variables) leads to a rise in R&D and innovations. A rise in R&D in this case
might also imply a rise in the net profit of each firm, for a given number of firms.
Proposition 6.1 below, which is a prediction on the joint effect of price competition on
innovations and market structure, would have to be modified to take that possibility
into account. If, however, a breakdown of collusion had no effect on R&D in the more
general model, proposition 6.1 should still hold unambiguously.
234 Chapter 6

function will often be convenient in the present analysis. Concentra-


tion C  , the number of firms N  , R&D expenditure R  , and innova-
tive output I  at the symmetric equilibrium are defined by the two
necessary conditions (6.1) and (6.2). Since I am here concerned with
comparative statics results, I will assume the existence of a unique
equilibrium with N  b 2 and R  > 0, I  > 0.
As in the case of advertising-intensive industries analyzed in
chapter 5, little can be said in general about the comparative statics
of I  , R  , and C  individually without imposing more structure on
the model. In particular, an intensification of price competition can
lead to more, less, or no change in R&D expenditure and innovative
output. This is because a rise in t can increase, decrease, or leave
unchanged the incentive to spend on R&D.
More specifically, a firm’s incentive to spend on R&D depends
on the total derivative dpi =dRi , and the effect of a rise in t on this
derivative is ambiguous. To see this, consider a decomposition of
the overall effect of a change in t on dpi =dRi based on writing pi ¼
ðpi  ci Þx i , where pi , ci and x i denote here, respectively, the equilib-
rium price, the equilibrium marginal cost, and the equilibrium output
of firm i in the final-stage subgame. The decomposition reveals four
different effects, two of which work in opposite directions for any
reasonable form of the profit function. To see this, note first that a
rise in Ri should generally cause both pi  ci and x i to increase, which
is why pi will also increase as a result. Now, on the one hand, the
higher the value of t, the higher the level of output sold, and hence
the larger the profit to be made by a unit increase in the price-cost
margin following a rise in Ri . This effect would imply that the higher
the value of t, the higher the value of the derivative dpi =dRi . On the
other hand, the higher the value of t, the lower the price-cost margin,
hence the smaller the profit to be made by a unit increase in output
following a rise in Ri . This second effect would imply that the higher
the value of t, the lower the value of the derivative dpi =dRi . There-
fore, provided that a rise in Ri causes both pi  ci and x i to increase,
as has been assumed here, the overall effect of a change in t on
dpi =dRi is ambiguous in general.
Not only is the effect of price competition on innovation ambig-
uous, but the effect on market structure can be uncertain. This is
because both sides of the free-entry condition will be affected by a
change in t, so it is not clear in what direction C  should change to
restore the free-entry condition.
Price Competition, Innovation, and Market Structure 235

The Joint Effect of Price Competition on Innovations and


Concentration

It is, however, possible to derive general results regarding the joint


behavior of I  and C  following a change in the intensity of price
competition t. The following proposition, which is analogous to
proposition 5.1 in chapter 5, excludes certain outcomes as not ad-
missible within the theory. In particular, the proposition provides a
sufficient condition for a rise in concentration following an increase
in t.

Proposition 6.1. If an increase in the intensity of price competition t


causes either an increase or no change in R  (and therefore also in I  ),
then concentration C  must rise.
Proof. It will be convenient to write gross profit directly as a func-
tion of the Ri ’s in what follows. From assumptions R1 and R3, we
know that pi is strictly concave in Ri . Moreover, from R2 and R3, pi is
nonincreasing in Rj , Ej 0 i. The gross profit of firm i at the initial
equilibrium can be written as
p0 ¼ pðRi ¼ R0 j S; C0 ; h; r; t0 ; Ri ¼ R0 Þ ¼ R0 þ f ;

and the gross profit at the new equilibrium is


p1 ¼ pðRi ¼ R1 j S; C1 ; h; r; t1 ; Ri ¼ R1 Þ ¼ R1 þ f :

Also, the profit of firm i at the initial equilibrium values of t and N 


when firm i sets Ri ¼ R1 while all other firms set Rj ¼ R0 is

p 0 ¼ pðRi ¼ R1 j S; C0 ; h; r; t0 ; Ri ¼ R0 Þ:


The proof is by contradiction. Assume that C1 a C0 and R1 b R0
following a rise of t from t0 to t1 . Since pi is decreasing in N, de-
creasing in t, and nonincreasing in Rj , Ej 0 i, we obtain p1 < p 0 . Also,
since the function pi ðRi j S; C0 ; h; r; t0 ; Ri ¼ R0 Þ is concave in Ri and its
slope is equal to 1 at Ri ¼ R0 , its slope is smaller than 1 at all points
between R0 and R1 . Hence p 0  p0 < R1  R0 . Combining the two
inequalities, we obtain p1  p0 < R1  R0 . This is, however, impos-
sible, since net profit must be zero at equilibrium, and therefore we
must have p1  p0 ¼ R1  R0 . Hence it must be the case that C  falls
if R  rises or does not change following a rise in t. r

The intuition for proposition 6.1 is as follows. Starting from a zero-


profit symmetric equilibrium, an increase in t unambiguously reduces
236 Chapter 6

net profit below zero, for given C and Ri ’s. For the zero-profit condi-
tion to be restored at a new symmetric equilibrium, net profit must
rise. Now suppose that the increase in t has also caused all firms to
spend more or exactly the same on R&D. This cannot increase net
profit because (1) an increase in own R&D has a stronger effect on
sunk cost than on gross profit when starting from a long-run equi-
librium with dpi =dRi ¼ 1, Ei (owing to the concavity of the gross
profit function with respect to Ri ), so it reduces gross profit; (2) if
own R&D does not change, it has no effect on profit; and (3) an in-
crease or no change in rival R&D cannot increase own profit. In these
circumstances, then, there is only one way net profit can increase,
through a rise in concentration.
Note that proposition 6.1 is a sufficient, but not necessary, condition
for a positive competition effect on concentration in R&D-intensive
industries. A positive competition effect on concentration can occur
even if the R&D expenditure and the innovative output of each firm
fall, provided that this fall is not too large and therefore does not
offset the direct negative effect of the rise in t on gross profit. Formal
theoretical results capturing this insight are easy to obtain, and they
are comparable to results obtained in section 5.2 of chapter 5 for
advertising-intensive industries. The key implication is that in R&D-
intensive industries, a nonpositive effect of competition on concen-
tration, although not impossible, may be unlikely.
Finally, note that proposition 6.1 is a prediction on the joint effect
of price competition on firm innovation and market structure. The
question arises as to whether anything can be said about the joint
effect of price competition on industry innovation and market struc-
ture. In particular, an interesting question is whether it is possible to
rule out the case DðN  I  Þ b 0 and DN  b 0 ðDC  a 0Þ, where DI  ,
DN  , and DC  denote the change in I  , N  , and C  , respectively. This
joint outcome can occur in one of two ways: either through DI  b 0
and DN  b 0; or through DI  < 0 and DN  b 0, with the rise in N 
offsetting the fall in I  , so that N  I  rises. The former case is ruled out
by proposition 6.1. The latter case cannot be ruled out on the basis of
the assumptions we have made, but it does seem very improbable.
To obtain DN  b 0 following a rise in t, R  needs to fall considerably;
if it falls by only a small amount, it will not offset the direct negative
effect of the rise in t on gross profit, so N  will fall. But since a sig-
nificant fall in R  is required for even a small rise (or no change) in
N  , it is very difficult to obtain DðN  I  Þ b 0 together with DN  b 0
ðDC  a 0Þ following a rise in t.
Price Competition, Innovation, and Market Structure 237

Discussion

The implications of the above model for the empirical analysis of


innovation at the industry level are, then, as follows. First, if indus-
try innovative activity increases or does not change following a rise
in the intensity of price competition, then it is very likely that con-
centration has risen. Second, if we actually observe that more price
competition causes a joint effect other than DðN  I  Þ b 0 and DC  a 0,
then proposition 6.1 will have been confirmed, since the joint out-
come DðN  I  Þ b 0 and DC  a 0 is a necessary condition for the joint
outcome DI  b 0 and DC  a 0. The empirical results on the evolution
of industry innovation and market structure in section 6.4 can there-
fore be used both to test proposition 6.1 and to examine whether it is
possible to establish an empirical regularity which implies a slightly
stronger constraint on the space of outcomes than proposition 6.1.
I will close this section with two brief remarks that draw on points
already raised in previous chapters. (I will omit, however, the usual
arguments regarding the exogeneity of the intensity of price compe-
tition t.) First, the assumption has been made throughout this section
that R&D and the entry decision are the only long-run choice vari-
ables in the game. This was motivated by the fact that the number of
industries where advertising and R&D are both key strategic vari-
ables is small. Hence it may be, on the whole, legitimate to assume
that the presence of endogenous advertising expenditures in some
R&D-intensive industries will not significantly alter the mechanism
and the results described in this section.10
Second, while the present results have been derived for the bench-
mark case of symmetric single-product firms, it is not clear what other
effects might arise in a more complex, asymmetric setting that would
offset the mechanisms identified here regarding the competition ef-
fect on innovation and market structure. On the other hand, there is
a potential complication with respect to the link between market size
and concentration. In particular, it is possible that the variance of
the firms’ growth rate within any given industry is bigger in R&D-
intensive industries than in other industries because of large asym-
metries between firms in exploiting profit opportunities through

10. An alternative approach would be to interpret R as the total amount of endoge-


nous sunk costs and I as an index of product quality or brand image. The results of this
section would then be valid for all classes of industries with significant endogenous
sunk costs (advertising or R&D).
238 Chapter 6

successful innovation.11 Now the larger this variance, the more likely
it is that an increase in market size will have a positive effect on
concentration, other things being equal. Of course, this is only one
potential mechanism through which market size may affect concen-
tration, and there are many others. The point is, however, that the
presence of potentially large asymmetries in the growth of firms in
R&D-intensive industries may cause a positive ‘‘bias’’ in the observed
market size effect on concentration relative to the benchmark predic-
tion for the symmetric case. Such a bias could be much more pro-
nounced in R&D-intensive industries than in other industries.
The theory should now be confronted with the empirical evidence.
The first question to be asked is what was the effect of the 1956 Act
on innovations in British industry. This is a question on which the
theory developed in this section is deliberately silent. Provided that
we observe a nonnegative effect, the next question will then be whether
concentration did indeed rise, as suggested by the present theory.

6.3 The Data

To analyze the effects of price competition in R&D-intensive indus-


tries, I will use several different samples of industries in separate
regressions for concentration and innovation counts. The reason for
using different samples is that the industry definitions across the
various data sources are often difficult to match. Many of the vari-
ables and the data sources are the same as those used in chapter 4.
This is the case for the concentration ratio, the setup cost proxies
(either the capital-labor ratio or capital divided by the number of
plants), the market size proxies (sales revenue deflated either by the
general producer price index or by industry-specific price indices),
and the competition variable CHANGE. The data used in the con-
centration regressions are, as in previous chapters, for 1958, 1963,
1968, and 1975.
The only new variable in this chapter is the number of innovations
produced by firms in each industry over time. Note that there is a
significant difference between innovations produced and innovations
used, and that I employ the first of these measures. This reflects the
fact that the focus of this chapter is on the determinants of innovative
output that is the result of own R&D, not the determinants of firms’

11. There is considerable evidence that the size distribution of the profit outcomes
from innovation is highly skewed. See, for instance, Scherer et al. (2000).
Price Competition, Innovation, and Market Structure 239

incentives to adopt innovations produced in other industries or by


other firms in the same industry.
The innovations database that I will use in this chapter is the only
available source of systematic information on innovative activity
across British manufacturing industries for the entire period exam-
ined in this book. Systematic data on other measures of innovative
output, such as patents, are not available prior to the 1960s. There
are, of course, well-known problems with measures of innovative
output such as innovation counts or patents. Attempts to count the
number of ‘‘significant’’ innovations are subject to some arbitrariness
and even possible biases in the evaluation procedure. Moreover,
even ‘‘significant’’ innovations differ in their economic value or the
R&D expenditure incurred. Patent counts are subject to even greater
problems, since patents are even more heterogeneous than significant
innovations with respect to their economic significance or the amount
of R&D involved. In addition, the propensity to patent innovations
varies considerably across industries. Despite their shortcomings,
however, measures of innovative output such as innovation counts
and patents have been used extensively in studies of innovation.
When panel data are available, as in the present case, many of these
shortcomings are much less serious, as will be discussed in detail
below.

Defining the samples of R&D-intensive industries

Although I will focus mostly on R&D-intensive industries, defined as


industries with an average or typical R&D-sales ratio (RDS) of more
than 1% over the relevant period, I also constructed several addi-
tional samples.
For the concentration regressions, I constructed a subsample of
high-R&D industries, defined as industries with average or typical
RDS higher than 2% over the relevant period. The objective was to
confirm that the results obtained from the full sample are not driven
by industries with medium R&D intensity (which may have features
of exogenous sunk cost industries). In other words, the idea was to
construct a subsample of industries for which R&D is clearly a key
strategic variable. The choice of the 2% cutoff point leaves a nonneg-
ligible number of industries with a change in competition regime in
the high-R&D subsample, and at the same time it ensures that this
subsample is sufficiently different from the RDS > 1% sample. In ad-
240 Chapter 6

dition, the 2% cutoff point splits the group of R&D-intensive indus-


tries used in the concentration regressions into two groups of roughly
equal size: of sixty-one industries with RDS > 1%, twenty-seven had
RDS > 2%. The classification of industries with respect to these cutoff
points was done in the manner described in chapter 4.
For innovation, it was not possible to use a similar subsample of
high-R&D industries, since there were very few industries with a
change in competition regime and average or typical RDS higher
than 2% over the period in the basic sample used in the innovation
regressions. In any case, given the results obtained from the inno-
vation regressions using the basic sample of industries with typi-
cal RDS > 1% (see below), it seemed more important to check the
robustness of these results using a larger group of industries than a
smaller one. To this end, I constructed a slightly enlarged sample,
including, in addition to those industries contained in the basic R&D-
intensive group, several industries with average RDS < 1% but with
a relatively high number of innovations in the database during the
period under study.12 This effectively amounts to using slightly modi-
fied criteria for selecting the group of technologically progressive
industries.

The Innovations Data: General Features

The innovations data used in this chapter come from a major survey
of significant innovations commercialized by UK firms during 1945–
1983. The survey was carried out by researchers in the Science Policy
Research Unit (SPRU) at the University of Sussex over a period of sev-
eral years and covers a total of 4,378 innovations. The data were col-
lected in three waves over a fifteen-year period: a first survey was
conducted in 1970 for the period 1945–1970 and identified 1,304 inno-
vations; a second, in 1980, covered the period 1970–1980 and identified
a further 848 innovations; and a third, carried out in 1983, covered
the whole period 1945–1983 and added another 2,226 innovations.

12. In particular, three groups of industries with typical RDS < 1% were initially
included in the enlarged sample: (1) industries in the nonferrous metal sector, whose
R&D-sales ratio was close to 1% for much of the period; (2) all industries with
RDS < 1% in the five core R&D-intensive sectors (chemicals, mechanical engineering,
instruments, electrical and electronic engineering, and vehicles); and (3) four more
industries with a large number of innovations in the database (iron and steel, glass
other than containers, paper, and plastic products). Some of these were later dropped
because of their ambiguous state of competition.
Price Competition, Innovation, and Market Structure 241

An innovation was defined, for the purposes of the SPRU survey,


as ‘‘the first successful commercial introduction of a significant new
product, process, material or system into British manufacturing in-
dustry or services.’’ This definition excludes mere inventions, orga-
nizational innovations, unsuccessful innovations, and the process of
diffusion. An explicit attempt was made by the researchers to ex-
clude all incremental innovations from the database.
The researchers asked nearly 400 industry experts to identify sig-
nificant technical innovations successfully commercialized in the UK
since 1945 in their sector of expertise, to name the firm or other orga-
nization responsible for each innovation, and to provide some basic
information for each innovation.13 The innovating firms were then
contacted, whenever possible, and asked to confirm the information
given by the experts and to provide additional and more detailed
information on a range of variables. Thus the database contains, for
each innovation, a brief description of the innovation; the date of
its introduction; the industrial classification and the ‘‘technical class’’
of the innovation; the name, size, and status of the ‘‘innovating unit’’
(for instance, whether an independent firm, a subsidiary, a public
body, etc.); the country of origin of the innovation; the industrial
classification of the innovating unit and of the first user of the inno-
vation; and information on a number of other variables. The SPRU
database is deposited in electronic form in the Data Archive at the
University of Essex.
The SPRU data have been extensively described and discussed by
Townsend et al. (1981), Robson et al. (1988), and Geroski (1994). More
than two-thirds of the total number of innovations are drawn from
five core sectors: chemicals, machinery and mechanical engineering,
instruments, electrical engineering and electronics, and vehicles. These
sectors are also the most R&D-intensive sectors in manufacturing.
Most of the remaining innovations come from four other sectors: basic
metals, shipbuilding and marine engineering, building materials, and
rubber and plastic goods. The remaining sectors innovate very little,
which is consistent with the fact that they spend very little on R&D.
Of the total number of entries in the database, about 75% are product
innovations and 25% are process innovations—although, as pointed

13. Each sector was covered by several experts who were drawn from research and
trade associations, academic institutions, government departments, trade and techni-
cal journals, individuals and consultants, as well as from firms.
242 Chapter 6

out in Robson et al. (1988), the definition of what is a process inno-


vation in an industry, rather than a product innovation by suppliers,
is often somewhat arbitrary. The prevalence of product innovations
in the SPRU data implies that most of these innovations were not
first used in the sector that produced them. An implication of this
is that a classification of innovations by user industry would differ
considerably from the classification by producer industry that I use
in this chapter.
There was a general upward trend in the number of innovations
introduced in the UK over 1945–1983, which corresponds to a similar
upward trend in the index of industrial production. Geroski (1994,
chapter 2) points out that the two series appear to be co-integrated
(i.e., that there seems to be a stable long-run relationship between
them). Nevertheless, the number of innovations in the data appears
to be relatively low relative to industrial production during the first
few years covered by the survey, and again during 1981–1983. In both
cases, this may be partly a consequence of the data collection proce-
dure.14 These potential inconsistencies in the data do not affect my
analysis, however, since neither the very early period nor the period
after 1980 is included in my data set of innovation counts for this
chapter.
One feature of the data that is worth mentioning, although it is not
directly relevant for the present study, is the large number of inno-
vations in the SPRU database that have been produced by small
firms.15 Moreover, it is worth bearing in mind that a large fraction of
the innovations in the database do not appear to have been the result
of very expensive R&D projects. This observation will turn out to be
useful when choosing an appropriate innovation lag for these data
(see below).

14. The low numbers of innovations during the early period may be partly due to the
difficulty of finding in 1970 experts who had been active during that period. And the
decline in innovation counts after 1980 may be due to the fact that this period was
covered in only one wave of data collection, while all the other years were covered in
two waves.
15. The share of innovations produced by larger firms is much lower than their share
of R&D expenditures, while the reverse is true for smaller firms (see Geroski 1994,
chapter 2). There may be various reasons for this, including underreporting of R&D by
small firms, overrepresentation of some low-concentration industries in the SPRU
database, failure to take into account links between small and large firms in the pro-
duction of innovations, and differences in the type of projects undertaken by larger
and smaller firms. Whatever the reasons, the SPRU data are certainly free from any
potential large-firm bias.
Price Competition, Innovation, and Market Structure 243

The Innovations Data: Problems and Solutions

As pointed out above, any attempt to count the number of significant


innovations is subject to some arbitrariness and possible biases in
the evaluation procedure. Fortunately, in the case of the SPRU data,
these biases seem to affect the comparability of innovative activity
across industries more than the comparability over time for any given
industry.
One feature of the data stands out in this respect. Although the
industries that contribute the most to the production of innovations
in the SPRU database are, on the whole, those that have high R&D
intensity, there is also considerable overrepresentation of some in-
dustries and underrepresentation of others. This clearly occurs to an
extent far greater than what can be justified by the fact that the
ranking of industries by R&D intensity is not necessarily the same as
the ranking of industries by innovative output (measured either by
innovation counts or by patents).16 The problem has been recognized
by the SPRU researchers themselves, and it is probably due to the
fact that experts in different sectors had different views on what con-
stitutes a ‘‘significant’’ technical innovation. However, these incon-
sistencies do not affect the analysis of the determinants of innovative
activity carried out in this book. This is because I use panel data, so
differences across industries in the measurement of innovative out-
put become part of the industry-specific effects and do not affect the
results of interest.
Although the most important inconsistencies in the SPRU data are
across industries, there are also some problems affecting the com-
parability of innovation counts over time in some industries. An
inspection of the raw data reveals that for several low-innovation
industries, there is overrepresentation of the period 1945–1970 and
underrepresentation of the period 1970–1980.17 On the other hand,
for at least two high-innovation industries, textile machinery (MLH
335) and scientific and industrial instruments (MLH 354), there is

16. Examples of industries with a disproportionately large number of innovations in


the database are textile machinery and mining machinery. Examples of industries with
a disproportionately small number of innovations in the data are toilet preparations,
polishes (both of which do not have even one innovation), and aerospace.
17. Examples include the entire sector of leather and leather goods, with more than
fifteen innovations during 1945–1970 and none thereafter, several food industries,
cement, and others.
244 Chapter 6

the opposite problem: overrepresentation of the period 1970–1980, as


is recognized by the SPRU researchers themselves (Townsend et al.
1981). These inconsistencies are probably due to the fact that the
data were collected in waves, so the same experts were not neces-
sarily contacted in each wave. None of the industries with obvious
time inconsistencies were included in the samples used for the in-
novation regressions of this chapter. Low-innovation industries
were not included anyway, since I have chosen to focus on R&D-
intensive or innovation-intensive industries only. In addition, I explic-
itly excluded textile machinery and instruments.
The inconsistencies that originate from the data collection proce-
dure are not the only problems with the SPRU data, however. Per-
haps the most important problem for my present purposes stems
from the way in which the innovations are assigned to the various
industries. In addition to a short description of each innovation and
an indication of its ‘‘technical class,’’ the data report the principal in-
dustry classification of the innovating firm or organization,18 the
industry classification of the innovation, and the industry classifi-
cation of the first user of the innovation. None of these is satisfactory
for my purposes.
The classification by first user industry is simply not relevant,
since I am interested in the production, not the use, of innovations.
The principal industry classification of the innovating firm may be
misleading in the case of diversified firms. Thus, for example, an in-
novation by the detergent division of a firm such as Unilever would
be classified in the food sector, since this is the principal industry
classification for Unilever. Since many firms in chemicals and engi-
neering (which form the bulk of the population of innovating firms)
are diversified, this is a significant problem. Finally, the industry clas-
sification of the innovation is misleading in the case of process inno-
vations, which, as pointed out above, are a considerable part of the
total in the SPRU data. For example, the industry classification of a
process innovation by the National Coal Board involving the use of
new or improved machinery would be ‘‘mining machinery,’’ although
this is clearly not an innovation by firms in the mining machinery
industry.
To minimize the measurement error from the misclassification of
innovations across industries, I reclassified all the innovations in the

18. In cases where the innovating unit is a subsidiary of another firm, the data report
the principal industry classification both of the subsidiary and the parent firm.
Price Competition, Innovation, and Market Structure 245

SPRU database according to the industry where each innovation was


produced. Broadly speaking, this involved classifying product in-
novations according to the industry of the innovation and process
innovations according to the industry in whose production process
the innovation was made. Thus, in the examples mentioned above, I
classified product innovations by the detergent division of Unilever
in the soap and detergent industry and process innovations involv-
ing new or improved mining machinery by the National Coal Board
in the coal industry. The reclassification was not always straightfor-
ward, however, and I often had to make use of the descriptions of
innovations given in the SPRU database or of independent informa-
tion regarding the industrial activities of specific firms.19 Since the
focus of the present study is on company-financed noncooperative
R&D, innovations by public research establishments and industrial
research associations were excluded; their number was small, any-
way. However, innovations by public firms were included.20

19. The most common difficulty was the following. Suppose that the principal indus-
try classification of the innovating firm/unit is A, the industry classification of the first
user of the innovation is again A, and the industry classification of the innovation is B.
It is then not always easy to tell whether this is a process innovation in industry A or a
product innovation in industry B by a division of that particular innovating firm/unit.
Clearly, in such a case it is helpful to know the exact description of the innovation
and whether the firm in question has a plant in industry B. Information on the latter
question for the year 1968 is contained in a detailed Directory of Business published as
part of the Census of Production of that year. Although this directory is obviously less
useful for innovations introduced in the 1950s than for those introduced in the 1960s
and early 1970s, and its coverage is not complete, it often provided the key informa-
tion required for classifying particular innovations. In other cases, my classification
was based on more subjective criteria. For instance, cases in which industry A in the
above example was iron and steel and industry B was instruments were classified as
process innovations, since it seemed unlikely that steel firms would have divisions
producing instruments. On the other hand, cases where both A and B were electronic
engineering industries were usually classified as product innovations, since firms in
that sector tend to be diversified into related industries.
20. There is also a second problem with the way the innovations are assigned to the
various industries in the SPRU database, which applies especially to low-innovation
sectors, such as food and drink, textiles, and clothing. In these sectors, the innovations
are very often assigned to the sector rather than to any particular industry. In fact, they
are assigned to the first ‘‘minimum list heading’’ (MLH) industry in each of these sec-
tors, which is, of course, very misleading. In most of these cases, it was possible to
reclassify the innovations to the appropriate industries. In any case, since my empirical
analysis focuses on R&D-intensive or innovation-intensive industries, the sectors that
were subject to serious measurement errors of this kind are not included in the sam-
ples used in this chapter. Note that MLH 411, man-made fibers, which is the first MLH
industry in the textile sector, is R&D-intensive. This industry was also dropped from
my samples, however, because of ambiguous state of competition in the 1950s.
246 Chapter 6

The level of aggregation used in the SPRU database is the three-


digit (or MLH) industry level, which is lower than the four-digit
industry level used in the concentration statistics. However, I was
often able to classify the innovations in the data according to more
disaggregated industry categories, roughly corresponding to the
four-digit industry level or to a level between the three-digit and the
four-digit. These were usually the MLH industry subdivisions in-
dicated in Central Statistical Office, Standard Industrial Classification
(London: H.M.S.O., 1968), although sometimes two or more such sub-
divisions were merged either to avoid creating industry categories
with very few innovations or because it was impossible to achieve a
very detailed industry breakdown of the innovation counts or of the
sales data. In other cases, it was not possible or not necessary to go
beyond the three-digit industry level, because the descriptions of the
innovations were not sufficiently detailed or the three-digit industry
was itself a relatively homogeneous market—both in terms of the
possible existence of R&D spillovers across products and from the
point of view of competition. One industry, aircraft, was excluded
because of heavy government involvement throughout the period.

Innovation Lags

Although the SPRU data record the number of innovations for every
year between 1945 and 1983, it is necessary to group the innovation
counts into somewhat longer periods, so that they can be matched
with Census-based figures on sales revenue and other variables that
are available only at roughly five-year intervals during the 1950s and
the 1960s. Using five-year periods to group the innovations is there-
fore an obvious choice, but this still leaves open the question of
exactly which periods to use.
Data on sales revenue, capital stock, employment, and so on are
available for 1954, 1958, 1963, 1968, and then annually from 1970 on-
ward. Data for 1953 are not available, but I constructed estimates for
1953 sales revenue on the assumption that the proportional change
of the deflated sales revenue in any given industry over 1953–1954
was equal to the average yearly rate of change of the deflated sales
revenue in that industry over 1954–1958. I used the same method to
construct estimates for the 1953 capital-labor ratio and the ratio of
capital stock to the number of plants in each industry. Now it is
reasonable to assume that the sales revenue in any given year t is a
measure of average market size between years t  2 and t þ 2. It is
Price Competition, Innovation, and Market Structure 247

also reasonable to assume that R&D expenditure in any given year is


influenced by market conditions (such as market size or the intensity
of price competition) in that year. It follows that the sales revenue in
year t can be thought of as a determinant of R&D spending between
years t  2 and t þ 2. The key question is What is the time lag be-
tween R&D spending and the commercialization of innovations?
It is well known that the time lag between the beginning of a re-
search project and the commercialization of innovations varies
greatly, depending on the nature and significance of the innovations.
Estimates of the average lag across industries will also depend on
the criteria used to compile any particular survey of innovations. In
fact, such estimates range from one to four years (see Mansfield et al.
1971, Pakes and Schankerman 1984, Acs and Audretsch 1988). In
addition, there is evidence, at least for the UK, that R&D expenditure
tends to increase in the last year or years of a research project (Schott
1976).
The simplest way of linking R&D spending to innovation counts in
a cross-industry study is to hypothesize that the bulk of R&D spend-
ing for innovations commercialized in year t takes place at a certain
year t  x, where x is an average across industries and types of in-
novations. This is the approach typically taken in the empirical liter-
ature on the determinants of innovative activity, and this is also the
approach that I will follow. What should be the value of x? If the
average lag between the beginning of a project and the commer-
cialization of innovations is one to four years, then the average lag
between the bulk of R&D spending and the commercialization of
innovations should be one to two years. On the assumption of a
one-year lag, the number of innovations commercialized in year t is
determined, on average, by market conditions in year t  1. Thus
sales revenue in year t (which, as already pointed out, is a measure
of average market size between years t  2 and t þ 2) should be
matched with the number of innovations commercialized between
years t  1 and t þ 3.21

21. An alternative way of linking R&D spending to innovation counts would be to


assume that the number of innovations in any given year t is a function of R&D
expenditures from year t  4 to year t, with appropriate weights applied for different
years. This approach is not very practical, however, when sales revenue data are
available only at five-year intervals. Besides, in the absence of any systematic infor-
mation on innovation lags in each particular industry, this approach is still only a
rough approximation to the actual innovation lags across industries and types of
innovations.
248 Chapter 6

The time periods chosen for grouping the innovations data in this
chapter are therefore 1952–1956, 1957–1961, 1962–1966, 1967–1971,
and 1972–1976. These were matched with data on sales revenue (and
capital intensity) for 1953 (my estimates), 1958, 1963, 1968, and 1973,
respectively. Note that these time periods are also very convenient in
the present case because focal-point years, such as 1955, 1960, 1965,
and so on are not borderline between any two consecutive periods.
This is important because the experts and the firms consulted for the
construction of the SPRU database may have used focal-point years
whenever they were uncertain about the exact date of the introduc-
tion of an innovation. The fact that focal-point years are close to the
midpoints of the time periods that I use here to group the innova-
tions should practically eliminate any errors resulting from imprecise
dating of innovations in the SPRU database.
These time periods are also convenient from the point of view of
the evolution of competition in British manufacturing. The number
of innovations commercialized in any given industry during 1952–
1956 was clearly determined by market conditions prior to the intro-
duction of the 1956 restrictive practices legislation. Innovative output
during the next period, 1957–1961, was driven by market conditions
during 1956–1960, which covers the last few years of cartelization as
well as the first two years after industries had started to formally
abandon price-fixing. But because of the fact that in many R&D-
intensive industries price-fixing agreements continued until the early
or even the mid-1960s, the period 1956–1960 can be seen, on the
whole, as a time when competition had not yet emerged to any sig-
nificant scale. The next period in the innovations data, 1962–1966, is
one in which we expect to see the short-run effect of competition on
innovation. In the final two periods, 1967–1971 and especially 1972–
1976, we almost certainly observe the long-run effect of competition.
While the assumption of a one-year innovation lag seems reason-
able and has been used in previous studies using the SPRU data (e.g.,
Blundell et al. 1995), I also experimented with an alternative set of
time periods for grouping innovation counts, chosen on the assump-
tion of an average two-year lag between the bulk of R&D spending
and the commercialization of innovations. This implied matching in-
novation periods 1953–1957, 1958–1962, 1963–1967, 1968–1972, and
1973–1977 with sales revenue data for 1953, 1958, 1963, 1968, and
1973, respectively. The choice of a two-year lag is consistent with the
evidence from a survey of R&D expenditure and innovations in a
Price Competition, Innovation, and Market Structure 249

sample of firms drawn among the 300 largest firms in British industry
in the early 1970s (Schott 1976). Note, however, that the study by
Schott focused on large or very large firms. On the other hand, a
large fraction of the innovations in the SPRU database are by small
firms, and the average R&D project of a small firm is likely to be less
expensive than that of a large firm. To the extent that this also implies
that smaller firms are generally involved in shorter-term projects than
larger firms, a one-year average innovation lag may be an appropriate
assumption for the SPRU data.
In any case, I chose to use the assumption of a one-year lag for
my basic specification of innovation counts (with dependent variable
INN1), but I also checked the robustness of the results under the
alternative assumption of a two-year average innovation lag (using
the alternative dependent variable INN2). Both sets of results will be
reported below, although they are, in fact, very similar.

Innovations Introduced or Innovations Produced?

A final complication with the SPRU data arises in connection with


the precise meaning of the term ‘‘introduction’’ of an innovation. In
particular, the introduction of an innovation in the UK by a domestic
‘‘innovating’’ firm (which can be the UK subsidiary of a multina-
tional firm) does not always imply that this firm actually produced the
innovation. A considerable number of innovations in the database
had been imported from outside the UK. Fortunately, the SPRU data
report the country of origin for each innovation. But, unfortunately,
there are serious doubts regarding the reliability of this information,
as pointed out by the SPRU researchers themselves (see Townsend et
al. 1981). The main problem seems to be the small number of inno-
vations that are reported as having originated outside the UK.22
Despite this difficulty, the country-of-origin indicator can be used
to check the robustness of the results to alternative definitions of
the dependent variable. If the exclusion of those innovations that are

22. Townsend et al. (1981) attribute this feature of the data to a variety of factors,
including a potential bias in the collected information. According to these authors, the
bias may have resulted from a mixture of technological nationalism, real difficulties in
distinguishing innovating firms and countries from imitating firms and countries, and
a high rate of import of foreign technology through mechanisms that are not easily
observable (such as imitation, reverse engineering, and inventing around patents), as
opposed to more measurable mechanisms (such as licensing, or transfers within mul-
tinational firms).
250 Chapter 6

reported as having originated outside the UK does not significantly


affect the results, one can safely conclude that the presence of im-
ported innovations in the database is not a serious problem. To check
this, I constructed the variable INN3, which is based on the popula-
tion of SPRU innovations reported as being of UK origin. In all other
respects, INN3 is similar to INN1, that is, it groups the innovations
into five-year periods assuming a one-year innovation lag. Bearing
this discussion in mind, I will use the terms ‘‘innovations produced’’
and ‘‘innovations introduced’’ interchangeably for the rest of this
chapter.

6.4 Empirical Models and Results

The econometric analysis of the effects of competition in R&D-


intensive industries in this section will be based on a comparison of
those industries affected by the 1956 Act with a control group of in-
dustries not affected. The theory developed in section 6.2 suggests
estimating the reduced-form models
Innovations ¼ IðS; f ; h; r; tÞ;

Concentration ¼ CðS; f ; h; r; tÞ;


where the variables are as defined in section 6.2. Time-invariant
industry-specific characteristics may also influence innovative activity
or concentration, and their effect will be captured in panel regressions
by the industry-specific effects. The two models will be estimated in
separate regressions—given that the two equations include the same
explanatory variables and it is also very difficult to match the in-
dustry definitions used in the concentration statistics with those used
for the innovations data.23 Once again, because of practical and con-
ceptual difficulties with estimating bounds, the analysis will be con-

23. It is also difficult to obtain any meaningful descriptive statistics on the joint evo-
lution of concentration and innovation counts in individual industries (such as those
reported for the joint evolution of advertising intensity and concentration in table 5.1).
There are two reasons for this. First, innovation counts tend to be volatile, so a rise or a
fall in innovation counts between the late 1950s and the late 1960s, say, in any given
industry may not in itself be very informative. Second, changes in innovative activity
are likely to be correlated with changes in sales, but it is not clear how to control for
this when focusing on the evolution of innovation counts in individual industries; for
instance, dividing the number of innovations by deflated sales revenue would give a
different picture than dividing it by the log of deflated sales revenue.
Price Competition, Innovation, and Market Structure 251

fined to obtaining regression results, on the assumption that the


theoretical predictions derived for the benchmark case of symmetric
single-product firms carry over to more complex settings.
In addition to the variables included in the above reduced forms,
other factors, including macroeconomic fluctuations, may have af-
fected innovations and concentration during the period examined in a
more or less uniform way across industries. As in previous chapters,
time dummies will be included among the regressors to control for
these factors. One variable that is not explicitly included in the the-
oretical model, but is generally thought to be an important deter-
minant of innovation, is the degree of appropriability of the outcome
of R&D; this is related to the degree of R&D spillovers within an
industry. This variable and two others, the degree of horizontal
product differentiation h and technological opportunity r, are very
difficult to measure accurately over time. However, it is not unrea-
sonable to assume that these variables will be relatively stable over a
period of ten or twenty years for the large majority of industries.
Hence they will be largely captured by the industry effects in the
panel data models estimated below.
As far as technological opportunity is concerned, it is well known
that the ranking of manufacturing sectors in terms of R&D intensity
tends to be very similar between countries and across time periods,
which is consistent with the view that technological opportunity is
relatively stable at the sector level. This is certainly less obvious for
three-digit industries and even less so for four-digit industries;
however, any changes in technological opportunity will only tend to
increase the ‘‘noise’’ in the results, as long as they are not correlated
with changes in the intensity of competition or other variables.

Price Competition and Innovation: The Samples and Descriptive


Statistics

As mentioned in section 6.3, I use two different samples of industries


to analyze the determinants of innovation in this chapter. Both are
balanced panels, with five periods for each industry. The first one
comprises all industries with an average or typical R&D-sales ratio
(RDS) of more than 1% over the period from the mid-1950s to the
mid-1970s, excluding industries with ambiguous state of competition
in the 1950s or later years. This sample contains thirty industries and
150 observations. The second sample includes, in addition, twelve
252 Chapter 6

Table 6.1
Innovation counts in 1952–1956, 1957–1961, 1962–1966, 1967–1971 and 1972–1976
Mean (st. deviation) of INN1
1952–1956 1957–1961 1962–1966 1967–1971 1972–1976

Industries with average RDS > 1%


Industries with 1.8 (2.2) 3.1 (2.7) 1.9 (2.3) 2.5 (2.1) 2.3 (1.1)
CHANGE ¼ 1
ðn ¼ 10Þ
Industries with 5.4 (4.9) 7.6 (7.8) 7.7 (6.0) 8.5 (8.3) 7.9 (8.0)
CHANGE ¼ 0
ðn ¼ 20Þ
Industries with average RDS > 1%
or a large number of innovations
in the SPRU database
Industries with 2.4 (2.7) 3.5 (3.1) 3.3 (3.5) 2.8 (2.1) 2.5 (1.5)
CHANGE ¼ 1
ðn ¼ 16Þ
Industries with 5.6 (4.6) 7.5 (7.3) 8.7 (7.2) 9.2 (8.6) 8.3 (7.7)
CHANGE ¼ 0
ðn ¼ 26Þ

Note: n indicates the number of industries.

industries with average RDS < 1% but with a relatively large num-
ber of innovations in the SPRU database. The full set of industries
and information on key variables are given in table B4 of appendix B.
Table 6.1 presents descriptive statistics for innovation counts sepa-
rately for industries with a change in competition regime and indus-
tries without such a change. The figures are based on the near-total
number of innovations introduced in British industry and recorded
in the SPRU database (i.e., excluding only a small number of inno-
vations produced by public research establishments or industrial
trade associations). Perhaps the most striking feature of the data is
the significant dissimilarity in initial conditions between the two
groups. In particular, the average number of innovations introduced
during 1952–1956 was 1.8 for R&D-intensive industries with a change
of competition regime in the 1960s and 5.4 for R&D-intensive indus-
tries without a subsequent change in regime (all of which were non-
collusive in the 1950s).
However, it would be very misleading to interpret this large dif-
ference in initial conditions as an indication of a negative effect of
collusion on innovative activity. An alternative interpretation, pro-
Price Competition, Innovation, and Market Structure 253

posed in chapter 3, is that collusive pricing may be more difficult


to achieve or to sustain in industries with high technological oppor-
tunity. Table 6.1 shows why this alternative interpretation is proba-
bly the correct one. The table compares the evolution of innovative
activity in the two groups of industries between the early 1950s and
the mid-1970s, when competition had been established in previously
collusive industries. Clearly, the large dissimilarity in the level of
innovative activity between the two groups persists throughout the
period, despite the trend toward more innovations in the average
industry over time. Furthermore, this trend is somewhat more pro-
nounced in industries without a change in competition regime, which
is certainly not consistent with the view that collusion has a negative
effect on innovation. On the other hand, the observed difference in
the evolution of innovative activity between the two groups is not
large in proportional terms, at least for the basic sample. Moreover, it
may be partly due to the fact that industries with CHANGE ¼ 1 were
growing more slowly than industries with CHANGE ¼ 0 during the
period under study. To examine these issues in a more powerful
way, I now turn to the econometric analysis of the determinants of
innovative activity.

Price Competition and Innovation: Econometric Specification

The basic model for innovation counts is


Innit ¼ f ða i ; b1 ln Salesit þ b2 lnðK=LÞit þ b3 Y58

þ b4 Y63 þ b 5 Y68 þ b6 Y73 þ b 7 CHANGE  Y58


þ b8 CHANGE  Y63 þ b9 CHANGE  Y68 þ b10 CHANGE  Y73Þ:

The dependent variable is the total number of innovations produced


by firms in any given industry and five-year period, defined in three
different ways, as described in the previous section. In particular,
INN1 is defined on the basis of the entire SPRU data set—excluding
only innovations by public research establishments or industrial
trade associations—and assuming a one-year lag between (the bulk
of ) R&D spending and the commercialization of innovations. Thus
the time periods used for grouping the innovations to construct
INN1 are 1952–1956, 1957–1961, 1962–1966, 1967–1971 and 1972–
1976. INN2 is defined on the basis of the same population of inno-
vations as INN1, but this time assuming a two-year innovation lag.
254 Chapter 6

In other words, the time periods for grouping the innovations are
in this case 1953–1957, 1958–1962, 1963–1967, 1968–1972 and 1973–
1977. Finally, to construct INN3 all innovations classified in the SPRU
data as having originated outside the UK were excluded, and the
innovations were grouped into five-year periods according to a one-
year innovation lag.
The independent variables are defined as follows. ‘‘Sales’’ is either
total sales revenue by UK firms deflated by the general producer
price index ðSSÞ or total sales revenue deflated by an industry-specific
producer price index ðDSÞ. The level of aggregation for these data is
the same as that for the innovation measures and the competition
variable CHANGE—that is, sometimes it is the three-digit industry
level, sometimes it is the four-digit industry level, and sometimes it
is between the three-digit and the four-digit industry level. K=L is the
capital-labor ratio, defined at the three-digit industry level.24 Y58,
Y63, Y68, and Y73 are time dummies corresponding, respectively, to
time periods 1957–1961, 1962–1966, 1967–1971 and 1972–1976 when
INN1 or INN3 is used, and to 1958–1962, 1963–1967, 1968–1972, and
1973–1977 when INN2 is used.
Finally, the interaction terms should reveal any differences in the
evolution of innovation counts since the mid-1950s between indus-
tries with a change in competition regime ðCHANGE ¼ 1Þ and indus-
tries without such a change ðCHANGE ¼ 0Þ. The benchmark period
is 1952–1956 (or 1953–1957), which is clearly before the introduc-
tion—let alone the implementation—of the legislation. Thus the
coefficient on CHANGE  Y58 measures the impact of the 1956 Act on
changes in innovative output between the benchmark period and the
period 1957–1961 (or 1958–1962); the coefficient on CHANGE  Y63
measures the impact of the 1956 Act on changes in innovative output
between the benchmark period and the period 1962–1966 (or 1963–
1967); and so on.
Four key features of the innovations data have directed the choice
of econometric specification. First, the dependent variable takes only
integer values, so a count data model must be used. Second, it is
evident from table 6.1 that there is very considerable overdispersion

24. I also experimented with the capital stock of the average plant, K/N, as a proxy for
setup cost. The results were similar to those reported here. Let me also repeat that
since the model includes industry-specific effects, one need not assume that K/L or K/N
is an accurate measure of setup cost. All that is required is that changes in K/L or K/N
are plausible measures of the change in setup cost.
Price Competition, Innovation, and Market Structure 255

in the innovations data; in other words, the variance is much larger


than the mean. The overdispersion is so pronounced that it will cer-
tainly persist in any Poisson regression: the conditional variance of
the dependent variable, although somewhat reduced through the in-
clusion of regressors, will remain larger than the conditional mean.
In such circumstances, the standard Poisson model is not appropri-
ate, even if one can partly control for heterogeneity through the in-
clusion of fixed industry-specific effects (see Cameron and Trivedi
1998, Hausman et al. 1984). On the other hand, both a random-effects
Poisson model and a negative binomial model can provide valid
results under overdispersion. Third, there is no ‘‘excess zeros’’ prob-
lem in the present data: only in 18 out of 150 observations in the basic
sample does the dependent variable INN1 take the value 0. And
fourth, as shown in table 6.1 and discussed in detail above, cartelized
industries had, on average, a much lower average number of innova-
tions throughout the period than noncartelized industries. This implies
that CHANGE is probably correlated with the industry-specific effects,
so a random-effects model cannot be used. To obtain consistent esti-
mates, the conditional fixed-effects negative binomial specification
proposed in Hausman et al. (1984) will be used below.25
A potential limitation of my specification for innovation counts is
that although it controls for fixed industry effects, it does not allow
for any effect of past innovative activity within an industry or by firms
in other industries on current innovation. As in previous chapters, a
dynamic specification cannot be used here because of data limitations.
However, the effect of past innovation on current innovation—unlike
the impact of industry-specific characteristics, which is captured by
the industry effects—may not be as important in industry-level data
as it often is in firm-level data. Moreover, while the production of
innovations in an industry may be generally influenced by knowl-
edge generated in other industries, Geroski (1994)—who worked
with annual firm-level and industry-level innovations data for much

25. This model is of the NB1 form, according to the terminology of Cameron and
Trivedi, that is, the conditional variance of the dependent variable is assumed to be a
multiple of the conditional mean. In addition to a multiplicative individual-specific
fixed effect, the model allows for an individual-specific overdispersion parameter (see
Cameron and Trivedi 1998). As pointed out by Blundell et al. (1995), this model
requires strict exogeneity of the regressors, an assumption which is not implausible in
the present context, where we use industry data and five-year periods. The dynamic
model proposed by Blundell et al. cannot be implemented here, since it requires a long
panel so that presample information can be used to approximate the fixed effects.
256 Chapter 6

of the period that I analyze in this book—found no evidence of sig-


nificant cross-industry (or within-industry) R&D spillovers in UK
manufacturing.

Price Competition and Innovation: The Identifying Assumption


Again

As in similar specifications used in previous chapters, the potential


endogeneity of CHANGE is an obvious cause for concern. Put sim-
ply, the potential problem is that any difference observed in the
evolution of innovation counts during the 1960s between industries
with CHANGE ¼ 1 and industries with CHANGE ¼ 0 may be to
some extent due to unobserved characteristics that differ between the
two groups of industries rather than to any effect of the 1956 cartel
legislation. Ideally, one would want to test formally for exogeneity,
but this is impossible here because there are no appropriate instru-
ments for CHANGE.
How serious is the problem? At first sight, it may seem rather se-
rious, since the initial conditions regarding innovative activity were
very different in the two groups. As shown in table 6.1, the mean
number of innovation counts during 1952–1956 was considerably
lower in industries with a subsequent change in competition regime
than in industries without such a change. However, table 6.1 also
shows that the picture was still very much the same ten or twenty
years later, when competition had generally been established in pre-
viously cartelized industries. This suggests that cartelization in the
1950s is correlated with some variable that strongly influences inno-
vative activity but remains relatively stable over time in any given
industry. An obvious candidate is technological opportunity. This—
as, indeed, any other time-invariant industry-specific characteristic—
is captured in the present model by the industry-specific effects.
Hence there should be no endogeneity bias, provided that one uses a
specification that accounts for the correlation between regressors and
industry effects.
There is an additional indirect check that can indicate whether the
potential endogeneity of CHANGE is likely to be a serious problem
in the data: one can compare the evolution of the two groups of in-
dustries between 1952–1956 and 1957–1962. The simplest way this
can be done is by examining the descriptive statistics reported in
table 6.1. Alternatively, one can check the regression coefficient on
Price Competition, Innovation, and Market Structure 257

CHANGE  Y58.26 Because (1) of the time lag between R&D spending
and the commercialization of innovations discussed in section 6.3 and
(2) of the fact that in several R&D-intensive industries price-fixing
agreements were not abandoned until the early 1960s or even later,
any changes in innovative activity between the period 1952–1956
and the period 1957–1962 cannot be attributed to the 1956 Act. If
there is an endogeneity problem with CHANGE that may affect the
regression results, then there should be evidence that the two groups
of industries evolved in different ways during the 1950s.
There is no such evidence from the descriptive statistics in table
6.1. In particular, the mean number of innovations produced in ten
R&D-intensive industries with a subsequent change in competition
regime was 1.8 during 1952–1956 and 3.1 during 1957–1961. The
respective figures for twenty R&D-intensive industries without a
change in regime were 5.4 and 7.6. In other words, there is no evi-
dence that innovative activity changed in different ways in the two
groups of industries during the 1950s: the increase in innovations
between 1952–1956 and 1957–1961 can be attributed to a common
time trend. As we will see below, this is confirmed by the regression
results: the coefficient on CHANGE  Y58 is not statistically significant.
We may therefore conclude that any estimated difference between
the two groups in later periods should be due to the 1956 legislation.
Alternatively, if it turns out that there is no difference between the
two groups in later periods, we may conclude that the legislation
had no effect on innovation.

Price Competition and Innovation: Econometric Results

Table 6.2 presents results for the basic sample of thirty R&D-intensive
industries (i.e., industries with average or typical RDS of more than
1% over the period from the mid-1950s to the mid-1970s). The sample
includes ten industries with a change of competition regime. The main
reason for focusing on R&D-intensive industries is the expectation
that any systematic impact of price competition on the production
of innovations may be more difficult to identify or less relevant in
industries where R&D is not a key strategic variable. Also, the theory
of section 6.2 relates specifically to R&D-intensive industries. In any

26. A valid objection to this second type of check is that if CHANGE were indeed
endogenous, then the coefficient on CHANGE  Y58 would itself be biased. However,
this check is intended here as a mere confirmation of the picture that emerges from the
descriptive statistics, not as a formal test.
258 Chapter 6

Table 6.2
Regression results for innovation counts in R&D-intensive industries (negative bino-
mial conditional fixed-effects estimation)
Dependent Dependent Dependent
variable: INN1 variable: INN2 variable: INN3
ln SS 0.27 — 0.19 — 0.25 —
(0.16) (0.16) (0.19)
ln DS — 0.17 — 0.11 — 0.23
(0.14) (0.14) (0.17)
ln K/L 0.15 0.07 0.07 0.03 0.28 0.22
(0.26) (0.26) (0.26) (0.26) (0.31) (0.31)
Y58 0.25 0.27 0.33 0.34 0.13 0.13
(0.16) (0.17) (0.17) (0.17) (0.20) (0.20)
Y63 0.19 0.25 0.19 0.23 0.13 0.14
(0.19) (0.19) (0.20) (0.20) (0.23) (0.23)
Y68 0.10 0.20 0.26 0.33 0.20 0.21
(0.26) (0.26) (0.25) (0.25) (0.29) (0.28)
Y73 0.11 0.04 0.14 0.23 0.02 0.01
(0.33) (0.33) (0.32) (0.32) (0.38) (0.37)
CHANGE  Y58 0.25 0.26 0.28 0.29 0.21 0.21
(0.36) (0.36) (0.37) (0.37) (0.41) (0.41)
CHANGE  Y63 0.26 0.28 0.11 0.12 0.40 0.41
(0.41) (0.40) (0.41) (0.41) (0.46) (0.45)
CHANGE  Y68 0.07 0.03 0.07 0.09 0.02 0.02
(0.38) (0.38) (0.39) (0.39) (0.41) (0.40)
CHANGE  Y73 0.18 0.11 0.07 0.02 0.02 0.03
(0.39) (0.39) (0.39) (0.39) (0.42) (0.42)
Constant 0.80 0.51 0.16 0.77 0.85 0.57
(1.96) (1.77) (2.01) (1.78) (2.36) (2.09)
Log likelihood 233.0 233.9 241.9 242.3 202.8 202.7
No. of industries 30 30 30 30 30 30
No. of industries 10 10 10 10 10 10
with CHANGE ¼ 1
No. of observations 150 150 150 150 150 150

Note: Standard errors in parentheses.


Price Competition, Innovation, and Market Structure 259

case, table 6.3 contains results for a larger sample of forty-two in-
dustries, including sixteen with a change of competition regime. This
sample includes, in addition to the R&D-intensive group, a number
of industries with average RDS < 1% but with a relatively large
number of innovations in the SPRU database.
There is no evidence of any overall effect of the intensification of
price competition, following the introduction of the 1956 Act, on
innovations introduced in British industry. The coefficients on the
interaction terms are not statistically significant, even at the 10%
or the 20% level, irrespective of the sample used or the form taken
by the dependent variable (INN1, INN2, or INN3). In particular, the
coefficient on CHANGE  Y63 is everywhere negative, but not sta-
tistically significant, while the coefficients on CHANGE  Y68 and
CHANGE  Y73 are sometimes positive and sometimes negative.
Clearly, the results provide no evidence of any short-run or long-run
effect of price competition on innovation in the present context.
An interesting feature of the results presented in tables 6.2 and 6.3
is the almost general failure of the explanatory variables in these
regressions. Time-invariant industry-specific characteristics seem to
account for much of the cross-industry variation in innovation counts.
As mentioned above, one variable that is almost certainly picked up
by the industry effects is technological opportunity. In addition, a lot
of variation in innovation counts seems to be due to variables diffi-
cult to measure or to observe, including random events.
It is somewhat reassuring that the coefficients on the market size
proxies, ln SS and ln DS, are often statistically significant at the 5%
or the 10% level in regressions using the larger sample (but not in
regressions using the basic sample of R&D-intensive industries). This
suggests that the larger amount of information contained in the
larger sample increases the efficiency of the estimated coefficients
on the market size variables. Note, however, that the magnitude of
these coefficients is fairly low, even in the larger sample. Taken at
face value, these estimates suggest that a 1% increase in market size
causes at most a 0.25% rise in the expected number of innovations.27

27. The rate of growth has sometimes been used instead of market size as a regressor
in empirical models of innovation. To see whether this makes any difference to my
results, I estimated a model including D ln SS or D ln DS among the regressors, defined
for industry i and year t as the change in ln SS or ln DS, respectively, in the five-year
period preceding year t. The coefficients on these variables were everywhere highly
nonsignificant, and the rest of the results did not change much. Note that in order to
use this alternative specification, the first-year observation for each industry had to be
dropped.
260 Chapter 6

Table 6.3
Regression results for innovation counts in the enlarged sample of innovative indus-
tries (negative binomial conditional fixed-effects estimation)
Dependent Dependent Dependent
variable: INN1 variable: INN2 variable: INN3
ln SS 0.27 — 0.19 — 0.25 —
(0.14) (0.14) (0.17)
ln DS — 0.18 — 0.13 — 0.25
(0.12) (0.12) (0.14)
ln K/L 0.19 0.13 0.10 0.06 0.17 0.12
(0.21) (0.21) (0.21) (0.21) (0.25) (0.25)
Y58 0.19 0.21 0.24 0.25 0.07 0.08
(0.14) (0.14) (0.14) (0.14) (0.17) (0.17)
Y63 0.22 0.28 0.22 0.26 0.18 0.18
(0.16) (0.16) (0.17) (0.16) (0.19) (0.19)
Y68 0.12 0.20 0.24 0.30 0.23 0.23
(0.21) (0.20) (0.21) (0.20) (0.24) (0.23)
Y73 0.12 0.01 0.06 0.13 0.03 0.04
(0.27) (0.26) (0.26) (0.25) (0.31) (0.29)
CHANGE  Y58 0.11 0.12 0.33 0.34 0.27 0.27
(0.27) (0.27) (0.27) (0.27) (0.30) (0.30)
CHANGE  Y63 0.05 0.07 0.03 0.05 0.07 0.07
(0.27) (0.27) (0.29) (0.29) (0.31) (0.31)
CHANGE  Y68 0.17 0.19 0.07 0.09 0.28 0.26
(0.28) (0.28) (0.28) (0.28) (0.31) (0.31)
CHANGE  Y73 0.12 0.15 0.16 0.17 0.15 0.12
(0.29) (0.29) (0.30) (0.30) (0.32) (0.32)
Constant 0.61 0.42 0.06 0.78 0.48 0.40
(1.67) (1.44) (1.69) (1.45) (1.98) (1.74)
Log likelihood 333.9 334.7 340.3 340.7 294.3 293.9
No. of industries 42 42 42 42 42 42
No. of industries 16 16 16 16 16 16
with CHANGE ¼ 1
No. of observations 210 210 210 210 210 210

Note: Standard errors in parentheses.


Price Competition, Innovation, and Market Structure 261

Price Competition and Concentration: The Samples and Descriptive


Statistics

I now turn to the analysis of the competition effect on concentration


in R&D-intensive industries. Recall that in the absence of any effect
of price competition on innovation, the theory predicts that concen-
tration should rise. Two samples of industries will be used for the
econometric analysis: a basic sample including all industries with an
average or typical R&D-sales ratio (RDS) of more than 1% over the
relevant period, and a subsample of industries with average or typi-
cal RDS higher than 2% over the relevant period. To construct these
samples, I started by excluding from the set of industries with avail-
able concentration data all industries with average or typical RDS
lower than 1%. Then I excluded industries with ambiguous state of
competition in 1958 (or, in a few cases, in the 1960s and early 1970s),
as well as industries with a switch of competition regime but for
which concentration data were not available for at least 1958, 1963,
and either 1968 or 1975. This left an unbalanced sample of sixty
R&D-intensive industries and 197 observations. To obtain the sub-
sample of high-R&D industries, I then also dropped all industries
with RDS < 2%; this left twenty-six industries and eighty-four ob-
servations. Table 6.4 gives details on the structure of these samples,
distinguishing between industries with a change in competition re-
gime after 1958 and industries without such a change. The list of
industries and information on key variables can be found in appen-
dix B (table B5).
Clearly, these samples are highly unbalanced, mainly as a result of
the fact that concentration data for 1958 are not available for a large
number of R&D-intensive industries. Although there is no reason
to suspect that this feature of the data might bias the results, it is
nevertheless worth checking this presumption using more balanced
samples. To this end, I also constructed a sample that includes only
industries with average or typical RDS higher than 1% and with a
full set of observations for at least the core period 1958–1968; in the
large majority of cases, the 1975 observation is also available.28 This
sample contains twenty-nine industries and 114 observations.
Some descriptive statistics on initial levels and changes in concen-
tration for R&D-intensive industries are reported in tables 6.5 and

28. I also included one industry (industrial engines) with available data for 1958, 1963,
and 1975, but not for 1968.
262 Chapter 6

Table 6.4
The basic samples for the concentration regressions: R&D-intensive industries and
high-R&D industries
No. of No. of No. of No. of
indus- indus- indus- indus-
tries with tries with tries with tries with
CHANGE ¼ 1 CHANGE ¼ 0 CHANGE ¼ 1 CHANGE ¼ 0
and average and average and average and average
Data available for RDS > 1% RDS > 1% RDS > 2% RDS > 2%
All four years 9 18 4 8
1958, 1963, 1968 0 1 0 0
1958, 1963, 1975 1 0 0 0
1963, 1968, 1975 0 22 0 8
1958, 1963 0 3 0 2
1963, 1968 0 6 0 4
Total 10 50 4 22

Table 6.5
Concentration in 1958 and competition regime in R&D-intensive and high-R&D
industries
Mean C5 in 1958
(st. deviation of C5)
Industries with average RDS > 1% ( basic sample)
Industries with CHANGE ¼ 1 ðn ¼ 10Þ 0.704 (0.171)
Industries with CHANGE ¼ 0 ðn ¼ 23Þ 0.683 (0.193)
Industries with average RDS > 1% (more balanced sample)
Industries with CHANGE ¼ 1 ðn ¼ 10Þ 0.704 (0.171)
Industries with CHANGE ¼ 0 ðn ¼ 19Þ 0.686 (0.196)
Industries with average RDS > 2% ( basic sample)
Industries with CHANGE ¼ 1 ðn ¼ 4Þ 0.699 (0.189)
Industries with CHANGE ¼ 0 ðn ¼ 10Þ 0.652 (0.246)
Industries with average RDS > 2% (more balanced sample)
Industries with CHANGE ¼ 1 ðn ¼ 4Þ 0.699 (0.189)
Industries with CHANGE ¼ 0 ðn ¼ 8Þ 0.639 (0.247)
Notes: The figures are based on industries with available data for 1958; n indicates the
number of industries.
Price Competition, Innovation, and Market Structure 263

Table 6.6
Average change in C5, in R&D-intensive and high-R&D industries, 1958–1968 and
1958–1975
DC5 DC5 DC5 DC5
1958–1968 1958–1975 1958–1968 1958–1975
Industries with average RDS > 1%
Industries with 0.093 0.122 0.093 0.118
CHANGE ¼ 1 (0.102) (0.111) (0.102) (0.118)
n¼9 n ¼ 10 n¼9 n¼9
Industries with 0.027 0.034 0.032 0.034
CHANGE ¼ 0 (0.125) (0.122) (0.127) (0.122)
n ¼ 19 n ¼ 18 n ¼ 18 n ¼ 18
Industries with average RDS > 2%
Industries with 0.152 0.111 0.152 0.111
CHANGE ¼ 1 (0.087) (0.096) (0.087) (0.096)
n¼4 n¼4 n¼4 n¼4
Industries with 0.025 0.042 0.025 0.042
CHANGE ¼ 0 (0.092) (0.110) (0.092) (0.110)
n¼8 n¼8 n¼8 n¼8
Notes: The figures in the first column are based on industries with available observa-
tions for both 1958 and 1968. The figures in the second column are based on industries
with available observations for both 1958 and 1975. The figures in the last two columns
are based on industries with available observations for 1958, 1968, and 1975. The fig-
ures in parentheses are standard deviations. n indicates the number of industries.

6.6. Table 6.5 reports descriptive statistics for the concentration ratio,
C5, in 1958 for industries with and industries without a change in
competition regime after 1958. There is no evidence of any significant
difference between the two groups: the mean C5 is 0.704 for the ten
cartelized industries in the sample and 0.683 for the twenty-three
noncartelized industries with available data for that year. On the
other hand, table 6.6 shows that the average increase in C5 after 1958
was much larger for R&D-intensive industries affected by the legis-
lation than for industries not affected. For the former group the
average change between 1958 and 1975 was about twelve percentage
points, while for the latter group it was only 3.4 percentage points.
A comparison of industries that experienced a large increase in C5
between 1958 and 1975 with those that experienced a large decrease
in C5 over the same period is also revealing. The comparison is limited
to the twenty-seven R&D-intensive industries with available data
264 Chapter 6

for both these years. Of the five industries with the largest rise in C5,
three (air and gas compressors; washing machines, electrically oper-
ated; electric lamps) are industries with CHANGE ¼ 1, and only two
(television receiving sets; radio communication equipment) have
CHANGE ¼ 0. However, all five industries with the largest fall in
C5 (cutlery; finished synthetic organic dyestuffs; powered industrial
trucks and industrial tractors; finished detergents; soap) have
CHANGE ¼ 0.
Of course, changes in other potential determinants of concentra-
tion may partly account for these differences. To control for these
other factors, we need to use econometric analysis.

Price Competition and Concentration: Econometric Model and Results

The basic econometric model for concentration is


Concit ¼ a i þ g1 ln Salesit þ g2 lnðK=xÞit þ g3 Y63 þ g4 Y68 þ g5 Y75
þ g6 CHANGE  Y63 þ g7 CHANGE  Y68

þ g8 CHANGE  Y75 þ eit ;


where ‘‘conc’’ is either the four-digit industry five-firm concentration
ratio, C5, or its logistic transformation, logit C5 ¼ ln½C5=ð1  C5Þ;
‘‘sales’’ is either SS or DS (see above); ‘‘K/x’’ is either the three-digit
industry capital-labor ratio, K=L, or the three-digit industry capital
stock of the average plant, K=N; Y63, Y68, and Y75 are time dummies
for 1963, 1968, and 1975 respectively; and the interaction terms cap-
ture any difference in the evolution of concentration after 1958 be-
tween industries with a change in competition regime and industries
without such a change. Thus the coefficient on CHANGE  Y63 mea-
sures the effect of the 1956 Act on concentration by 1963, the coefficient
on CHANGE  Y68 measures the effect by 1968, and the coefficient on
CHANGE  Y75 measures the effect by 1975. The benchmark year is
1958. CHANGE is defined according to the industry categories used
for C5. Once again, the potential endogeneity of some of the inde-
pendent variables may be an issue here. The response to this is the
same as in previous chapters.
Results for concentration were obtained using both a fixed-effects
model and a random-effects model. The two sets of results were very
similar, except for the coefficients on the setup cost proxies, which
were highly significant in the random-effects model only. Since the
Hausman test invariably and clearly failed to reject the random-
Price Competition, Innovation, and Market Structure 265

effects model and—unlike the case of the innovation regressions


above—there are no a priori reasons for doubting the validity of the
hypothesis that the regressors are not correlated with the industry
effects, the results from the random-effects model are presented be-
low. I also report, however, some of the results derived using a fixed-
effects specification to facilitate comparisons with previous chapters.
Once again, heteroskedasticity was present in these regressions, and
it was more pronounced in regressions using logit C5 than in those us-
ing C5. Although the reported standard errors are heteroskedasticity-
consistent, the results from the specifications using logit C5 may be
subject to the objection that too much weight is placed on a few
industries with very high concentration.
Tables 6.7 and 6.8 present results for the basic sample of R&D-
intensive industries, while table 6.9 focuses on the more balanced
sample of R&D-intensive industries with available observations for
at least the three core years 1958, 1963, and 1968. There are ten in-
dustries with canceled agreements in these samples. All the results
provide clear evidence of a strong positive competition effect on con-
centration in R&D-intensive industries. In particular, the coefficients
on CHANGE  Y68 and CHANGE  Y75 are everywhere positive and
statistically significant at the 5% level. According to these coefficients,
the introduction of cartel law caused, on average, a rise in the five-
firm concentration ratio of as much as ten percentage points in R&D-
intensive industries. This effect was largely realized between 1963
and 1968. On the other hand, there was very little effect between
1958 and 1963 in this class of industries. It should be noted, however,
that these samples include three industries where agreements were
not abandoned until after 1963, and in two of these competition did
not fully emerge until after 1968.29 Finally, note that the market size
effect on concentration breaks down in R&D-intensive industries, as
expected: the coefficients on the sales variables have conflicting signs
and they are nowhere significant at the 5% level.
These results are not driven by industries with medium R&D in-
tensity. To check that the competition effect persists at high levels of
R&D intensity, I also estimated the model using the sample of in-
dustries with typical RDS > 2% over the period; the results appear
in table 6.10. These results should be interpreted with some caution,

29. They also include an industry with a change of regime but for which C5 is not
available for 1968 although it is available for 1975 (so CHANGE  75 picks up all the
effect of the 1956 Act between 1963 and 1975 in this industry).
Table 6.7

266
Regression results for concentration in industries with average RDS > 1% (random-effects estimation)
Dependent variable: C5 Dependent variable: logit C5
ln SS 0.010 — 0.006 — 0.15 — 0.23 —
(0.015) (0.016) (0.13) (0.14)
ln DS — 0.006 — 0.002 — 0.20 — 0.27
(0.014) (0.014) (0.13) (0.14)
ln K/N 0.073 0.073 — — 0.66 0.66 — —
(0.016) (0.016) (0.15) (0.15)
ln K/L — — 0.062 0.063 — — 0.61 0.62
(0.019) (0.019) (0.19) (0.19)
Y63 0.014 0.015 0.009 0.011 0.15 0.16 0.15 0.16
(0.022) (0.022) (0.022) (0.022) (0.17) (0.17) (0.17) (0.17)
Y68 0.016 0.019 0.008 0.011 0.24 0.28 0.24 0.29
(0.025) (0.025) (0.026) (0.026) (0.20) (0.20) (0.21) (0.22)
Y75 0.024 0.027 0.017 0.020 0.38 0.44 0.41 0.48
(0.029) (0.029) (0.031) (0.031) (0.21) (0.22) (0.24) (0.25)
CHANGE  Y63 0.003 0.002 0.002 0.002 0.06 0.05 0.12 0.10
(0.035) (0.035) (0.035) (0.035) (0.24) (0.24) (0.23) (0.23)
CHANGE  Y68 0.091 0.092 0.084 0.085 1.10 1.10 1.09 1.09
(0.032) (0.032) (0.033) (0.033) (0.71) (0.71) (0.71) (0.71)

Chapter 6
CHANGE  Y75 0.112 0.113 0.104 0.105 0.90 0.91 0.90 0.91
(0.031) (0.031) (0.034) (0.033) (0.32) (0.32) (0.33) (0.32)
Constant 0.786 0.740 0.675 0.634 0.60 1.04 2.08 2.47

Price Competition, Innovation, and Market Structure


(0.152) (0.137) (0.161) (0.144) (1.34) (1.30) (1.49) (1.46)
Hausman statistic 7.88 5.50 4.55 4.06 7.30 — 2.33 2.89
Prob-value 0.45 0.70 0.80 0.85 0.50 0.97 0.94
R2 0.29 0.29 0.19 0.20 0.31 0.31 0.22 0.22
No. of industries 60 60 60 60 60 60 60 60
No. of industries 10 10 10 10 10 10 10 10
with CHANGE ¼ 1
No. of observations 197 197 197 197 197 197 197 197
Note: Heteroskedasticity-consistent standard errors in parentheses.

267
Table 6.8

268
Regression results for concentration in industries with average RDS > 1% (fixed-effects estimation)
Dependent variable: C5 Dependent variable: logit C5
ln SS 0.019 — 0.019 — 0.27 — 0.29 —
(0.023) (0.022) (0.27) (0.27)
ln DS — 0.011 — 0.011 — 0.36 — 0.36
(0.020) (0.020) (0.24) (0.24)
ln K/N 0.030 0.032 — — 0.17 0.14 — —
(0.029) (0.029) (0.29) (0.30)
ln K/L — — 0.018 0.022 — — 0.25 0.21
(0.035) (0.036) (0.41) (0.40)
Y63 0.001 0.002 0.005 0.002 0.07 0.10 0.07 0.10
(0.026) (0.025) (0.025) (0.024) (0.21) (0.21) (0.22) (0.21)
Y68 0.016 0.011 0.025 0.019 0.02 0.09 0.05 0.11
(0.033) (0.032) (0.031) (0.031) (0.29) (0.29) (0.34) (0.33)
Y75 0.019 0.014 0.030 0.022 0.07 0.17 0.13 0.21
(0.040) (0.039) (0.039) (0.038) (0.35) (0.35) (0.44) (0.42)
CHANGE  Y63 0.008 0.007 0.009 0.008 0.01 0.02 0.01 0.02
(0.035) (0.035) (0.035) (0.035) (0.29) (0.29) (0.29) (0.29)
CHANGE  Y68 0.074 0.076 0.069 0.071 0.97 0.96 0.95 0.95
(0.040) (0.040) (0.040) (0.040) (0.64) (0.63) (0.65) (0.64)

Chapter 6
CHANGE  Y75 0.093 0.096 0.087 0.090 0.80 0.82 0.78 0.81
(0.040) (0.040) (0.041) (0.040) (0.43) (0.43) (0.42) (0.42)
R2 0.23 0.22 0.22 0.22 0.15 0.16 0.15 0.16

Price Competition, Innovation, and Market Structure


R 2LSDV 0.92 0.92 0.92 0.92 0.82 0.83 0.82 0.83
No. of industries 60 60 60 60 60 60 60 60
No. of industries 10 10 10 10 10 10 10 10
with CHANGE ¼ 1
No. of observations 197 197 197 197 197 197 197 197
Note: Heteroskedasticity-consistent standard errors in parentheses.

269
Table 6.9

270
Regression results for concentration in industries with average RDS > 1%: more balanced sample (random-effects estimation)
Dependent variable: C5 Dependent variable: logit C5
ln SS 0.007 — 0.004 — 0.18 — 0.29 —
(0.019) (0.019) (0.18) (0.18)
ln DS — 0.002 — 0.001 — 0.21 — 0.30
(0.017) (0.017) (0.16) (0.16)
ln K/N 0.071 0.071 — — 0.67 0.67 — —
(0.023) (0.023) (0.19) (0.19)
ln K/L — — 0.064 0.065 — — 0.60 0.59
(0.035) (0.035) (0.31) (0.31)
Y63 0.001 0.003 0.003 0.001 0.08 0.10 0.09 0.11
(0.026) (0.026) (0.026) (0.026) (0.21) (0.21) (0.21) (0.21)
Y68 0.009 0.013 0.003 0.006 0.17 0.20 0.17 0.20
(0.030) (0.030) (0.032) (0.032) (0.24) (0.24) (0.27) (0.27)
Y75 0.021 0.025 0.018 0.022 0.51 0.57 0.56 0.61
(0.038) (0.038) (0.044) (0.043) (0.29) (0.30) (0.38) (0.39)
CHANGE  Y63 0.013 0.013 0.012 0.012 0.04 0.04 0.03 0.03
(0.037) (0.038) (0.037) (0.037) (0.27) (0.27) (0.25) (0.25)
CHANGE  Y68 0.082 0.082 0.076 0.076 0.99 0.99 1.00 0.99
(0.035) (0.035) (0.036) (0.036) (0.74) (0.74) (0.73) (0.73)

Chapter 6
CHANGE  Y75 0.106 0.108 0.101 0.103 0.99 1.00 1.02 1.03
(0.036) (0.036) (0.038) (0.038) (0.34) (0.34) (0.35) (0.35)
Constant 0.772 0.718 0.667 0.618 0.86 1.10 2.61 2.72

Price Competition, Innovation, and Market Structure


(0.192) (0.169) (0.196) (0.174) (1.79) (1.62) (1.85) (1.68)
Hausman statistic 3.79 3.63 2.29 2.11 6.60 7.40 3.82 4.70
Prob-value 0.88 0.89 0.97 0.98 0.58 0.49 0.87 0.79
R2 0.30 0.30 0.18 0.18 0.32 0.31 0.20 0.19
No. of industries 29 29 29 29 29 29 29 29
No. of industries 10 10 10 10 10 10 10 10
with CHANGE ¼ 1
No. of observations 114 114 114 114 114 114 114 114
Note: Heteroskedasticity-consistent standard errors in parentheses.

271
Table 6.10

272
Regression results for concentration in industries with average RDS > 2% (random-effects estimation)
Dependent variable: C5 Dependent variable: logit C5
ln SS 0.006 — 0.011 — 0.16 — 0.39 —
(0.018) (0.020) (0.18) (0.18)
ln DS — 0.013 — 0.017 — 0.27 — 0.47
(0.016) (0.018) (0.16) (0.18)
ln K/N 0.053 0.053 — — 0.81 0.80 — —
(0.029) (0.029) (0.27) (0.27)
ln K/L — — 0.044 0.044 — — 0.76 0.78
(0.038) (0.037) (0.36) (0.36)
Y63 0.024 0.028 0.025 0.029 0.24 0.30 0.37 0.44
(0.032) (0.032) (0.034) (0.033) (0.31) (0.31) (0.33) (0.33)
Y68 0.010 0.017 0.006 0.013 0.31 0.44 0.45 0.60
(0.039) (0.039) (0.042) (0.041) (0.35) (0.36) (0.40) (0.41)
Y75 0.023 0.034 0.021 0.032 0.58 0.77 0.85 1.08
(0.052) (0.053) (0.061) (0.061) (0.40) (0.41) (0.48) (0.51)
CHANGE  Y63 0.0001 0.001 0.002 0.002 0.14 0.13 0.26 0.25
(0.055) (0.055) (0.055) (0.056) (0.38) (0.38) (0.39) (0.39)
CHANGE  Y68 0.144 0.146 0.134 0.135 2.29 2.29 2.25 2.26
(0.041) (0.041) (0.041) (0.041) (1.45) (1.44) (1.42) (1.41)

Chapter 6
CHANGE  Y75 0.111 0.117 0.101 0.106 0.96 1.04 1.04 1.12
(0.056) (0.056) (0.058) (0.057) (0.56) (0.56) (0.57) (0.55)
Constant 0.608 0.546 0.517 0.465 0.82 1.89 3.79 4.61

Price Competition, Innovation, and Market Structure


(0.183) (0.161) (0.219) (0.197) (1.79) (1.63) (1.87) (1.92)
Hausman statistic 6.01 6.02 2.26 2.38 8.64 9.85 7.98 1.55
Prob-value 0.65 0.64 0.97 0.97 0.37 0.28 0.44 0.99
R2 0.28 0.25 0.14 0.13 0.34 0.32 0.19 0.18
No. of industries 26 26 26 26 26 26 26 26
No. of industries 4 4 4 4 4 4 4 4
with CHANGE ¼ 1
No. of observations 84 84 84 84 84 84 84 84
Note: Heteroskedasticity-consistent standard errors in parentheses.

273
274 Chapter 6

since the number of industries with a change in competition regime


in this sample is only four, all of which are drawn from the electrical
engineering sector.30 The main reason for reporting these results is to
show that there is no indication of any weakening of the competition
effect on concentration at relatively high levels of R&D intensity.

6.5 Concluding Remarks

This chapter has analyzed the impact of price competition on inno-


vative output and market structure in R&D-intensive industries. The
theory suggests that strong predictions on the effect of competition
on innovation are difficult to derive in general. On the other hand, it
is possible to derive a prediction on the evolution of market structure
that is conditional on the behavior of innovative activity following a
rise in the intensity of price competition. In particular, if innovative
activity does not fall with more price competition, then concentration
must rise. In fact, a positive effect of price competition on concen-
tration is the most likely outcome in R&D-intensive industries, since
the reverse can occur only in the event of a significant fall in R&D
expenditure.
The introduction of the 1956 Restrictive Trade Practices Act pro-
vides a rare opportunity for addressing these issues in an empirical
context. An econometric analysis based on a comparison between
those industries affected by the legislation and those not affected has
produced no evidence of any significant overall effect of price com-
petition on innovation at the industry level, and clear evidence of a
strong positive overall effect on concentration in R&D-intensive in-
dustries. The effect on concentration was at least as large as that
estimated for exogenous sunk cost industries. The empirical evidence
is therefore consistent with the theory.
The empirical results of this chapter are based on reduced-form
econometric equations in which concentration and innovation are
treated as endogenous variables, while the intensity of price compe-
tition is taken as exogenous. The results are therefore subject to the
usual limitations of this type of econometric analysis. The use of
reduced-form equations is a powerful way of revealing overall effects,

30. The very high coefficients on CHANGE  Y68 in these regressions may be partly
due to ‘‘overshooting’’ of the concentration ratio in electrical engineering industries
affected by a series of mergers in the second half of the 1960s, including the 1967–1968
mergers between A.E.I., G.E.C., and English Electric, three of the largest and most
diversified UK electrical engineering firms.
Price Competition, Innovation, and Market Structure 275

but it may fail to capture some potentially interesting interactions


between the variables of interest. For instance, it could be the case
that innovations are often introduced by new entrants (as argued by
Geroski 1994). An increase in the intensity of price competition could
then cause innovations to fall by reducing new entry, and thus the
pool of potentially important innovators. At the same time, the in-
tensification of price competition could induce existing firms to in-
novate more in order to avoid bankruptcy and exit. Hence a zero
overall effect of competition on innovation could be consistent with a
more complex mechanism than the one identified in this chapter. Or
it could be the case, as suggested by recent theoretical results on the
competition-innovation relationship based on theoretical models of
endogenous growth (Aghion and Howitt 1997, 1998), that competi-
tion reduces innovation under certain circumstances and increases it
under different circumstances. Again, this could be consistent with
the zero overall effect suggested by our analysis.31
These various possible mechanisms and theoretical predictions
from more specific models (or others that focus on potential inter-
actions between innovation and market structure) can be tested only
with data far richer and more detailed than those used in this book.
But, in any case, our inability to test these hypotheses in the present
context does not alter the basic result of this chapter, namely, the
lack of any overall effect of price competition on innovation and the
presence of a strong competition effect on concentration across R&D-
intensive industries following the abolition of restrictive agreements
in British manufacturing in the 1960s. The result on concentration
confirms similar results in previous chapters. On the other hand, the
result on innovations, which is consistent with the Broadberry and
Crafts (2000) finding of no significant effect of collusive pricing on

31. A third possibility might be that high concentration leads to low (high) innovation,
everything else being equal, so that an increase in the intensity of competition, by
causing concentration to rise, leads to a fall (rise) in innovations. At the same time, the
intensification of price competition could induce firms to innovate more (less) at any
given level of concentration. Again the overall effect could be zero, but the interaction
of the various factors would be more complex than what is implied by the present
analysis.
However, this interpretation does not seem to be consistent with the facts. Because
market structure adjusted only gradually in R&D-intensive industries affected by the
1956 Act, as shown in the concentration regressions of this chapter, some insight on the
interaction of innovation and market structure can be gained by comparing the short-
run effect with the long-run effect of the 1956 Act. The results of tables 6.8 and 6.9,
taken as a whole, provide little evidence that the short-run effect of the Act (i.e., the
effect before the adjustment of market structure) was different from the long-run effect.
276 Chapter 6

innovation in the 1950s, has interesting implications for the debate


on the links between price and nonprice competition.
In particular, we saw in chapter 5 that stronger price competition
has probably reduced advertising expenditure in British industry.
The results of the present chapter indicate that the production of
innovations has been largely independent of the intensity of price
competition. Thus there seems to be a fundamental difference be-
tween advertising and R&D, which has not been adequately ana-
lyzed in existing theories of competition. There are, of course, many
obvious and important differences between advertising and R&D.
The question is Why would any of these differences imply that price
competition should generally have a negative effect on advertising
but an ambiguous effect on R&D? This question is outside our pres-
ent scope, and it is not clear whether it can be fully addressed within
the present theory. Nevertheless, one remark may be appropriate here,
since it builds on a distinction between advertising and R&D made
in chapter 3.
It was pointed out in that chapter that a low-quality firm in an
R&D-intensive industry may find it difficult to compete with a high-
quality rival, while in an advertising-intensive industry heavy adver-
tisers may face significant competition from firms that advertise little
and employ different marketing strategies. Under certain conditions,
this will imply that firms may be more inclined to cut back advertis-
ing expenditure than R&D expenditure when faced with a change in
the competitive environment that causes a fall in gross profit. Now
even if that were true, we would still need to explain why this be-
havior should persist in the long run, and it is here that a potential
limitation of the present theory may be coming into focus. Expressed
in somewhat general terms, the limitation may be as follows: If his-
tory matters, and there is a multiplicity of possible long-run out-
comes, then the comparative statics of the long-run equilibria may
sometimes depend on the short-run dynamics. In other words, there
may be instances where systematic short-run effects persist in the
long run and influence the long-run equilibria. The study of these
short-run effects might then help narrow down the space of possible
(or likely) long-run outcomes.
Chapter 7 returns to one of the strongest mechanisms analyzed in
this book. Unlike the present chapter and the two previous ones, it
adopts a more unifying approach in order to provide more detail and
a richer understanding of the basic mechanism that links competition,
market structure, and profitability across all classes of industries.
7 Price Competition and
Profitability: Are Cartel
Laws Bad for Business?

7.1 Introduction

Recent advances in oligopoly theory have shed new light on the


much-debated issue of the links between firms’ conduct, market
structure, and market performance. Selten (1984) and Sutton (1991,
1998) have shown that under free entry and exit, an intensification of
price competition in a homogeneous-good industry will have no sig-
nificant effect on firms’ profits in the long run, although it will cause a
decrease in the number of firms. Previous chapters of the present book
have described and extended the theory of the competition-market
structure relationship to advertising-intensive and R&D-intensive
industries, and have also provided several empirical tests of the
theory for different classes of industries. The empirical results were
indeed consistent with the Selten-Sutton predictions for exogenous
sunk cost industries. Moreover, the results suggested that the mech-
anism that causes concentration to rise when price competition in-
tensifies is so strong that it will only infrequently break down in
advertising-intensive or R&D-intensive industries.
There is, however, one aspect of the present theory that has not yet
been empirically examined, and that is the effect of competition on
profits. This chapter concludes the empirical analysis of the effects of
the 1956 Restrictive Trade Practices Act in this book by examining the
joint effect of price competition on market structure and profitability.
The only previous statistical analysis of the effects of the 1956 Act
on profitability is the study by O’Brien et al. (1979). These authors
used firm-level data taken from company accounts for a sample of
about thirty industries and examined two different measures of prof-
itability: the rate of return on capital and the rate of return on sales.
They found a decline in the mean level of profitability between the
278 Chapter 7

period 1951–1958 and the period 1958–1967 for their full sample of
firms, but could not identify any significant difference in the evolu-
tion of profitability between firms in industries affected by the 1956
Act and those in industries not affected. They also found weak evi-
dence of a recovery of the rate of return on capital over the period
1968–1972 in industries affected by the restrictive practices legisla-
tion relative to the control group of noncollusive industries, although
this was not statistically significant. Moreover, a recovery of the rate
of return on capital was also observed for firms in industries with
agreements upheld by the Restrictive Practices Court relative to firms
in the control group. Thus the results provided no evidence of any
effect of the 1956 Act on the profitability of firms. O’Brien et al. (1979)
also examined merger activity over the same period, but again found
no evidence of any significant difference between industries with a
change in competition regime and those without such a change. As
mentioned in chapter 4 of this book, the main methodological limi-
tations of the O’Brien et al. study were the use of a rather small sample
of industries and the fact that the criteria for classifying industries
across groups were somewhat dubious in a few cases.
Despite these limitations, the O’Brien et al. study is one of a few
empirical analyses of the links between competition and profitability
that has explicitly treated concentration as an endogenous variable and
not as a proxy for the degree of market power. The literature on the
effects of economic integration on industry performance has some-
times followed a similar approach, using import penetration or the
degree of tariff protection as measures of competitive pressure and
recognizing the potential effect of changes in these variables on mar-
ket structure as well as on performance. Many of these studies find a
negative effect of import competition on price-cost margins, espe-
cially in concentrated industries, which they typically interpret as
being consistent with the hypothesis that foreign competition reduces
the market power of domestic firms (see, for instance, Jacquemin et al.
1980, deMelo and Urata 1986). There are, however, several important
methodological differences between these studies and the present one.
First, some of the profit measures that are relevant in the context of
the present theory have not been generally used in the existing liter-
ature on the determinants of profitability. Second, economic integra-
tion leads to changes not only in firms’ conduct but also in product
substitutability and market size. Hence it is not easy to draw very
clear implications for the competition-profitability relationship from
Price Competition and Profitability 279

studies of the effects of import competition on industry performance.


Third, and most important, none of these studies has examined the
interaction between changes in market structure and changes in
profitability or compared the short-run and the long-run evolution of
profitability following an intensification of competition. These are,
however, key issues within the theory developed in this book. As a
result, the econometric approach adopted here differs in important
respects from that of previous studies. In particular, since the theory’s
predictions regarding the long-run competition effect on profitability
depend on allowing market structure to change to restore the long-
run equilibrium, it is important for testing these predictions that one
does not control for changes in market structure when specifying the
profitability equation.
In the remainder of this chapter, I begin with a brief reminder and
clarification of the theoretical predictions regarding the impact of
price competition on profitability. This is followed by a brief discus-
sion of some case-study evidence from industries affected by the
1956 Act. In particular, I describe two case studies that seem to con-
firm the theoretical results, then present a counterexample that illus-
trates the limitations of the present theory and yet at the same time
can be seen as a sharp test of the theory’s predictive power.
Finally, I provide econometric results on the joint effect of com-
petition on market structure and profitability. My market structure
measures in this chapter are the number of firms and the number of
plants; this analysis therefore complements the one carried out in
previous chapters using the concentration ratio. The main advantage
of using firm or plant numbers is that data on these come from ex-
actly the same sources as the profit data. Therefore the samples used
for the market structure regressions and the profit regressions in this
chapter are identical. I use three different profit measures: the gross
profit of the average firm, the gross profit of the average plant, and
the industry price-cost margin. My results are in line with the theory,
justifying the recent shift of emphasis toward models that endoge-
nize market structure and consequently emphasize the effect of firm
conduct on structure rather than on performance.

7.2 Theoretical Issues

The mechanism that I will test in this chapter has been described in
previous chapters, so I will provide only a brief summary and some
280 Chapter 7

extensions and clarifications in what follows. Let me start again with


the benchmark case. Under free entry, the net profit of each of a
number of symmetric firms must be zero—or almost zero, taking
account of the integer constraint and assuming that the number of
firms is not very small—irrespective of firm conduct. An increase
in the intensity of price competition, caused by the introduction of
cartel laws, economic integration, or some other exogenous institu-
tional change, will cause gross profit and the price-cost margin to
fall, given the initial number of firms. As a result, firms will no longer
be able to cover their sunk costs at the initial free-entry equilib-
rium. In an exogenous sunk cost industry, this will inevitably lead to
mergers and exit until the gross profit of each firm rises sufficiently
to cover its sunk costs, which have remained unchanged. Note that
the fact that the capital stock is already in place when competition
intensifies is not important in the long run because capital depreci-
ates and has to be renewed. Thus the degree to which setup costs
are sunk when competition intensifies and the rate of depreciation of
the capital stock will affect only the speed of adjustment to the new
long-run equilibrium market structure.
At the new long-run equilibrium, the number of firms (or plants)
will be lower, but there will be no significant change in firm (or plant)
gross or net profit. More specifically, one would expect profitability
to decline in the short run, that is, before any significant change in
market structure occurs, and then be restored, or partially restored,
in the long run through a fall in firm numbers. This is the basic
mechanism that the present chapter sets out to test.
If the integer constraint is taken into account, little can be said
about the exact effect of more intense price competition on profit
without imposing more structure on the model. In the specific model
analyzed by Selten (1984), both total industry net profit and plant (or
firm) net profit are more likely to increase than to decrease following
a switch from a collusive to a noncollusive regime (see also Phlips
1995, chapter 3). However, in a context where the integer constraint
is the only reason for positive net profit, this can be expected to be
small in general, at least at the plant or firm level, so any change
in profit may be difficult to identify empirically. Since, in addition,
there is no clear prediction as to the direction of the change, it seems
legitimate to consider a weaker version of the Selten result as the
main testable prediction of the theory, namely, that a switch of com-
Price Competition and Profitability 281

petition regime has no significant effect on plant (or firm) gross or net
profit in the long run.1

Two Key Assumptions

Clearly, these results depend on two crucial assumptions. The first is


that incumbent firms under a collusive regime cannot prevent entry
of new firms into the industry or, more generally, that the scope for
entry deterrence is not larger in a collusive equilibrium than in a
noncollusive one. If this were not the case, then a breakdown of
collusion would reduce net and gross profit and would have an
ambiguous effect on the number of firms.
It is, in fact, possible to construct theoretical models of collusion
where firms make supranormal profits either by adopting trigger
strategies that deter entry (Harrington 1989, 1991) or by adopting a
strategy of gradually accommodating entry (Friedman and Thisse
1994). In such models, a decrease in the degree of collusion will re-
duce net and gross profit and will have an ambiguous effect on the
number of firms—assuming that the scope for entry deterrence is
smaller in a more competitive equilibrium. Hence an empirical test of
the present theory regarding the effect of competition on profits may
also be interpreted as a test of alternative theories of collusion.
The second key assumption is the symmetry assumption. How-
ever, the typical industry is subject to significant asymmetries, due
to a variety of factors, including multiplant or multiproduct firms,
or efficiency differences between firms. In previous chapters I have
argued that the presence of asymmetries is unlikely to modify the
theoretical predictions regarding the effect of price competition on
market structure in any significant way. This may not be true for the
effect on profitability. In the presence of asymmetries, free entry is
consistent with supranormal profits for all but the marginal firm in an
industry, and it is not clear whether any general theoretical predic-
tion can be derived about the long-run equilibrium effect of more
intense price competition on the profit of the average firm.
In fact, results from specific models suggest that the effect of
tougher price competition on firm profit and even on the price-cost
margin may be positive in the presence of efficiency differences, even

1. Also, the effect on industry net profit will be ambiguous, although industry gross
profit should fall because of the fall in the number of firms.
282 Chapter 7

when no account is taken of the integer constraint (see Montagna


1995). The intuition is that an exogenous shock that reduces prices in
the short run drives the less efficient firms out of the industry, so in
long-run equilibrium price may fall by less than the marginal cost of
the average firm in the industry, and thus the price-cost margin of
the average firm may rise. In other words, low-cost firms will expand
at the expense of high-cost rivals, and since low-cost firms always
have higher margins in an asymmetric industry, the possibility arises
that the margin of the average firm will be higher in the new long-
run equilibrium despite the increase in the intensity of competition.
Of course, this is only a possibility, and it could also be the case that
the overall effect is a fall in the margin of the average firm. For my
present purposes, it is sufficient to note that in the presence of effi-
ciency differences, the long-run effect of price competition on profit-
ability is ambiguous.
The above discussion suggests that the present theory can accom-
modate several of the predictions of models that emphasize efficiency
differences between firms. However, it is at odds with one potential
implication of some of these models, namely, that the number of firms
need not fall in the long run when price competition intensifies. The
reason why this could happen is as follows.2 An intensification of
price competition would put pressure on price-cost margins and
cause low-cost firms to expand at the expense of high-cost firms in
the short run; this redistribution of market shares would also raise
concentration. Because of the expansion effect, it is possible, if the
efficiency differences are sufficiently large, that the gross profit of
low-cost firms increases given the initial number of firms. This would
lead to entry of low-cost firms. Of course, the gross profit of high-
cost firms would decrease, thus causing exit of those firms. But this
whole mechanism opens up the possibility that the total number of
firms does not fall (or, at any rate, does not fall significantly) in the
long run, following an increase in the intensity of price competition,
even though concentration should rise as a result of the redistribu-
tion of market shares among firms. It also opens up the possibility
that industry profitability does not fluctuate (i.e., fall in the short run
and then rise following changes in market structure). This mecha-
nism, which could be regarded perhaps as a strong version of the

2. See Aghion and Schankerman (2000), where this possibility is discussed along with
many others that are not inconsistent with the predictions of the present theory.
Price Competition and Profitability 283

efficiency differences approach, is clearly incompatible with the


theory developed in this book. The empirical evidence presented in
this chapter can therefore also be seen as a test of the two competing
models.

Endogenous Sunk Costs

I have so far focused on the case of an exogenous sunk cost industry.


Things can be more complicated in industries with significant endog-
enous sunk costs, such as advertising or R&D. A change in the inten-
sity of price competition will affect the incentive of each firm to spend
on advertising or R&D, which is part of the total sunk cost incurred.
As a result, it may not always be the case that more intense price
competition causes the number of firms to fall. On the other hand,
the long-run effect on net profit will still be (approximately) zero,
although gross profit may change because of the change in advertis-
ing/R&D expenditure, and the direction of that change is not pre-
dictable in general.
As shown in previous chapters, however, the competition-concen-
tration relationship will only infrequently break down in practice in
advertising-intensive or R&D-intensive industries. Thus one should
not overemphasize the potential differences across classes of indus-
tries in this respect. The same may be true for the effect of price com-
petition on profitability. It is not difficult to show that under plausible
assumptions, this effect should be negative in the short run (i.e., before
any change in market structure occurs) and can be ambiguous in the
long run (i.e., once market structure adjusts) in industries with endog-
enous sunk costs, just as in exogenous sunk cost industries. To see
this, consider two cases.
Suppose, first, that price competition has no significant effect on
the nonprice variable at the industry level, as was the case in chapter
6 with innovative output and (presumably) R&D expenditure. Gross
profit and the price-cost margin must then fall in the short run when
price competition intensifies, given that neither R&D nor market
structure has changed. In the long run, both gross profit and the
price-cost margin can rise or fall, depending on market structure,
total industry sales revenue, and the type and degree of asymmetries
between firms. For instance, if firms are symmetric and their number
falls and industry sales revenue rises in the long run following an
284 Chapter 7

intensification of price competition, the free-entry condition implies


that the effect on the price-cost margin can be ambiguous.
Next, suppose that price competition has a negative effect on the
nonprice variable at the industry level, as was the case in chapter 5
with advertising expenditure or intensity. In the short run, gross
profit and the price-cost margin should decline as a result of more
intense price competition—and this effect could be reinforced by the
fall in advertising. In the long run, the effect is ambiguous. For in-
stance, if firms are symmetric and their number falls, advertising
expenditure or intensity at the firm level can rise or fall, and the free-
entry condition then implies that the same is true for profitability.

Summary of the Theory and Implications for the Empirical Analysis

Two main conclusions can be drawn from the above discussion. First,
in the presence of asymmetries between firms or endogenous sunk
costs, such as advertising or R&D, it is difficult to derive any strong
theoretical results regarding the effect of price competition on prof-
itability in the long run. Second, the basic intuition from the bench-
mark case—that profitability should initially decline following a rise
in the intensity of price competition and then be restored, or par-
tially restored, through a change in market structure—emerges as
the dominant mechanism driving the joint evolution of structure and
performance across classes of industries. The only exception to this
can occur in rather special cases where price competition has a par-
ticularly strong effect on nonprice variables; however, the evidence
from previous chapters suggests that there are very few such cases
in the present context. This has two implications for the empirical
analysis of profitability in this chapter.
The first implication is that the key testable prediction in this
chapter is that profitability should initially decline following an in-
tensification of price competition, and then be restored through a fall
in firm numbers. This will be tested against alternative predictions
derived from two other theories mentioned above: the theory that
cartels deter entry and a strong version of the efficiency differences
approach.
The second implication is that there is probably little to be gained
from performing separate regressions for each class of industries. On
the other hand, there is an important gain from pooling the data,
namely, an increase in the efficiency of the estimation. This is impor-
Price Competition and Profitability 285

tant here for several reasons. First, the effect of price competition on
firm or plant numbers may be less easy to identify than the effect
on concentration, because firm or plant numbers are very sensitive
to the number of small firms or plants, and these may have not been
affected by the 1956 Act to the same extent as larger firms or plants
(see section 7.4). Second, there are few advertising-intensive or R&D-
intensive industries with a change in competition regime in the
present sample. Given that profit data are inherently ‘‘noisy,’’ it is
questionable whether any meaningful results can be derived in sep-
arate regressions for advertising-intensive and R&D-intensive indus-
tries. Third, since one of the key testable predictions in this chapter
is that price competition has no significant long-run effect on profit,
one should use the largest possible sample to give the data the best
possible chance to reveal a nonzero effect. Moreover, another key
aspect of the analysis will be the comparison of short-run and long-
run effects, and again the difficulty of doing this may increase as the
size of the sample used decreases.
In any case, since the theoretical predictions for the effect of price
competition on profit are perhaps less clear for advertising-intensive
and R&D-intensive industries than for exogenous sunk cost indus-
tries, I have also run regressions for the subsample of exogenous
sunk cost industries; the results will be discussed below. The main
purpose of this is to check that the results for the whole sample are
not driven by advertising-intensive or R&D-intensive industries, be-
cause that would not be consistent with the theory.
Before presenting the econometric results, however, I will discuss
some case-study evidence that illustrates very nicely the central
structural mechanism that drives the results of this chapter and of
the book as a whole. In fact, part of the interest of these case studies
derives from the fact that they can confirm that the observed patterns
in the cross-industry data identified in this chapter and in previous
chapters are the result of the mechanism proposed in the theory
rather than of other factors quite different from those suggested by
the theory.

7.3 Case-Study Evidence

Several of the case studies of the effects of the 1956 Restrictive Trade
Practices Act discussed in Swann et al. (1973) provide evidence that
is consistent with the theory described in section 7.2. Admittedly, the
286 Chapter 7

evidence is somewhat sketchy, since the main focus of these case


studies is on the effect of the Act on competition—and not on its
repercussions on market structure and industry performance. More-
over, it is not always easy to separate the effect of competition from
the influence of other factors on profitability and market structure in
the industries investigated. Nevertheless, a brief description of the
evolution of some of the industries in the Swann et al. sample be-
tween the late 1950s and the early 1970s should help provide a better
understanding of the mechanisms that drive the econometric results
obtained later in this chapter. Two of the best-documented cases are
the transformer industry and the glass container industry.

The Electrical Power Transformer Industry

The details on the operation of collusive arrangements in the electrical


power transformer industry during the 1950s became widely known
even before the implementation of the 1956 Act, since the industry
was the subject of a Monopolies and Restrictive Practices Commis-
sion inquiry in the mid-1950s, along with several other subdivisions
of the electrical machinery industry (see MRPC 1957). Subsequently,
the industry registered its agreements and defended the most im-
portant of these in the Restrictive Practices Court, which struck it
down in early 1961.
The agreement covered the home market and related to all but the
smallest sizes of transformers. It provided for common minimum net
selling prices, the reporting of inquiries and orders received from
customers, and aggregated rebates (i.e., discounts to buyers on the
basis of the total quantity purchased from firms who were parties
to the agreement). These firms were members of the Transformer
Makers’ Association, which had always included all the producers of
very large transformers. However, there was some competition from
a few outside firms for large, as opposed to very large, transformers,
and significant competition from outside firms for smaller equipment.
It is therefore not surprising that the Monopolies and Restrictive
Practices Commission found that the profitability of association firms
was lower for the smaller sizes of transformers than for the larger
sizes. On the whole, profitability was higher than the average for all
manufacturing, but not excessive, according to the Commission. The
membership of the association declined somewhat in the late 1950s,
Price Competition and Profitability 287

so outside competition may have been stronger in 1961 than at the


time of the Commission’s investigation.
Collusion among members of the association had been effective
during the 1950s, and competition was slow to emerge after the
agreement was formally abandoned following the Court judgment.
Explicit price-fixing was replaced by the exchange of information on
prices and tenders, and this helped sustain prices and margins for
several years, despite occasional price-cutting (see Swann et al. 1973).
Interestingly, several firms that were not members of the association
at the time of the Court hearing seem to have adhered to the implicit
arrangements after 1961.
Eventually, however, competition was triggered by a combination
of events, the most important of which may have been the significant
fall in demand after 1964. A second factor was the decision of the
Restrictive Practices Court to strike down two information agree-
ments in other industries on the grounds that they had amounted
to the same effect as explicit price-fixing (see chapter 2). A third fac-
tor was a more aggressive purchasing policy on the part of certain
buyers. As a result, prices fell by as much as 25–40% between 1964
and 1968, according to figures reported by Swann et al., while costs
slightly increased during the same period. The price fall was more
pronounced for medium-sized transformers than for very large
equipment.
The main response of the industry to this substantial decline
in prices and profits was merger and exit. The evolution of market
structure in transformers was, of course, closely linked with devel-
opments in other markets for electrical equipment, all of which ex-
perienced an intensification of price competition during the 1960s.
But nowhere was the change in market structure so pronounced as
in transformers. According to figures based on the Census of Pro-
duction, the five-firm sales concentration ratio in the industry, which
had fallen from 50% in 1958 to 45% in 1963, jumped to 76.7% in 1968,
an increase of over thirty percentage points in five years. A large part
of this increase was due to mergers: Swann et al. (1973) list more
than ten mergers between producers of transformers (and other
electrical machinery) over the period 1965–1969.3 Admittedly, this

3. Unfortunately, the Census figures on firm numbers in the transformer industry are
not comparable over the period 1963–1968, because of changes in the definition of
subproducts within the industry.
288 Chapter 7

was probably a case of overshooting, as suggested by the fact that


the five-firm concentration ratio declined slightly after 1968, reaching
68.8% in 1975.
By the late 1960s, then, the industry was much more concentrated
and the number of firms was much smaller than in the early 1960s or
even in 1964, the date when prices started falling. Did prices rise in
the late 1960s, as would be predicted by the theory of this chapter?
The answer is that they did—indeed, quite sharply. The evidence
reported by Swann et al. suggests that prices started to rise in 1969
and were higher in 1971 than in 1963. In fact, price rises of up to 50%
were reported for 1969–1970 and 1970–1971. This was attributed
by industry experts to the elimination of excess capacity through
mergers and exit.

The Glass Container Industry

The experience of the glass container industry has much in common


with that of the electrical power transformer industry. As in the case
of transformers, collusion had a long history in the glass container
industry, and had been facilitated during the 1950s by the steady and
moderate growth in demand. The agreement among producers of
glass containers came before the Restrictive Practices Court in 1961
and was defended by the parties. Under the agreement, the members
of the British Bottle Association, who were responsible for about
80% of the total UK production of glass containers, had to observe
common minimum prices and standard conditions of sale, and also
offered aggregated rebates to distributors. Competition from outside
producers of glass containers was weak, since several nonmembers
sold only to specific buyers and/or tacitly observed the association
prices. However, there was also competition from producers of con-
tainers of different materials, such as paper, tin, and plastic, and this
was on the increase at the time the case was heard in the Court.
Swann et al. (1973) report that price competition emerged in the in-
dustry soon after the explicit price-fixing agreement was condemned
by the Court, despite an attempt to sustain collusion through the op-
eration of an information agreement. Unlike the transformer industry,
however, there were no spectacular price falls in glass containers,
possibly because demand fluctuations were not very pronounced.
Moreover, the price falls that occurred are somewhat difficult to in-
Price Competition and Profitability 289

terpret because the cost of materials seems also to have fallen during
the first half of the 1960s. The clearest indication of a change in
competition regime is therefore the decline in profitability in the
industry in the first half of the 1960s, as documented in Swann et al.
on the basis of data from company reports.
In particular, the largest firm in the industry experienced a con-
tinuous decline in its rate of return to capital, which fell from 18% in
1960 to 5.8% in 1965. For the other firms the evidence is somewhat
mixed, but for all of them profitability clearly declined over 1962–
1964 (i.e., in the first few years after the emergence of price competi-
tion). However, profitability recovered in the mid-1960s. And while
the evidence provided by Swann et al. on the profitability of various
firms in the second half of the 1960s is rather sketchy, the overall
picture is one of sustained moderate profitability for the industry
as a whole. At the same time, prices in early 1969 were at roughly
the same level as in late 1965, although wages and costs of materials
had been rising during that period. Why did the rate of return on
capital recover in the mid-1960s, and how was it then sustained de-
spite increases in input prices that were not fully passed on to selling
prices?
The answer is probably twofold. First, technical progress in the
industry led to increased efficiency during the 1960s and helped to
keep unit costs low despite increases in input prices. Second, the re-
covery and subsequent stability of profits must have been related to
the restructuring of the industry, which was partly realized through
a series of mergers beginning in 1962 and continuing until the late
1960s. Thus, according to highly disaggregated data from the Census
of Production, the number of producers (excluding those with fewer
than twenty-five employees) declined by as much as 50% in all the
principal product lines within the glass container industry between
1958 and 1968, while the five-firm sales concentration ratio for the
industry as a whole increased from about 63.5% in 1958 to 69.5% in
1963, and to 87.3% in 1968, a rise of nearly twenty percentage points
in five years.
The two case studies briefly described above provide support for
the theoretical predictions of the present chapter. Are they typical
examples of the effect of cartel policy on profits and market structure?
This question can be answered only by using econometric analysis.
Prior to that, however, it is worth discussing a counterexample.
290 Chapter 7

The Secondary Battery Industry

In the case of the secondary battery industry, the predictions of this


chapter are not confirmed by the facts. The reason is that a key as-
sumption of the theory, the assumption of free entry under the collu-
sive regime, is violated. In one sense, then, this case study illustrates
the limitations of the theory. In another sense, and perhaps somewhat
paradoxically, it confirms the present theory, albeit in a different way
than the two previous cases. In particular, the evolution of the sec-
ondary battery industry between the late 1950s and the early 1970s
demonstrates the relevance of a structural interpretation of the effects
of the 1956 Act on profits and market structure, as opposed to a purely
behavioral interpretation.
The secondary battery industry was investigated by the Mono-
polies Commission in the early 1960s as part of the Commission’s
inquiry on various classes of electrical equipment for vehicles (see
Monopolies Commission 1963). Detailed information on the evolu-
tion of the industry in the 1950s and the 1960s can therefore be drawn
both from the Commission’s report and from Swann et al. (1973).
Sales of secondary batteries fall into one of four main types: auto-
motive batteries sold as initial equipment, automotive batteries sold
for replacement, traction batteries sold as initial equipment, and
traction batteries sold for replacement. The structure of the industry
has always been characterized by a relatively high degree of con-
centration, especially in the initial equipment market, which was
dominated throughout the 1950s and the 1960s by two firms, Lucas
and Chloride. In the case of automotive replacement batteries, sev-
eral smaller firms also were active, and their market share increased
during the second half of the 1950s, as we will see below. Overall, the
combined market share of Lucas and Chloride was around 70% of
total industry sales revenue in 1960.
These two firms were the most prominent members of the British
Starter Battery Association, which also included two smaller firms. A
number of other firms remained outside the association, either be-
cause they chose to do so or because they were refused membership.
The agreement between the members related to replacement bat-
teries for motor vehicles and provided for common list prices, com-
mon discounts for various classes of buyers, and restrictions on the
relationships between manufacturers and buyers. The regulations
Price Competition and Profitability 291

affecting the distribution of batteries are of considerable interest for


our present purposes.
There were several classes of buyers, but a key distinction was
between ‘‘service agents’’ and other types of buyers. Each of the four
association members had its own network of service agents, who
could buy only from that particular manufacturer. Their names, ad-
dresses, and total number had to be reported to the association, and
no transfer of an agent from one manufacturer to another could take
place without the consent of both manufacturers concerned. In ad-
dition, the member firms could sell directly to wholesalers, retailers,
or fleet operators. There were no individual exclusive dealing arrange-
ments for these classes of buyers, but there was collective exclusive
dealing. In other words, these buyers were required to deal only with
association firms if they were to be supplied at all.
In addition to the agreement on replacement batteries, there was a
tacit understanding between Lucas and Chloride that they would not
canvass one another’s customers in the automotive initial equipment
market. As for traction batteries, there were occasional discussions
on prices among three firms, including Chloride, which were the
only significant producers of this type of battery.
The arrangements on initial equipment and traction batteries were
terminated when the 1956 Act was introduced, and were never regis-
tered. On the other hand, the agreement of the British Starter Battery
Association was revised and then registered. The main changes were
the abandonment of common list prices and the abolition of collec-
tive exclusive dealing; clearly, the purpose of the changes was to make
the agreement more acceptable to the Court. The revised agreement
provided for maximum trade discounts and retained the individual
exclusive dealing arrangements with ‘‘service agents.’’ According to
Swann et al., the abolition of collective exclusive dealing had a sig-
nificant impact on the structure of the industry, since smaller firms
that had previously found it difficult to enter the industry or expand
due to the lack of distribution outlets could now emerge as legitimate
competitors in the replacement market. As a result, the combined
market share of nonassociation firms in the replacement battery mar-
ket doubled within a few years to reach approximately 50% in the
mid-1960s.
In 1960 this revised agreement was replaced by an information
agreement. Soon after that, the Monopolies Commission began its
292 Chapter 7

investigation of the industry. The principal conclusions of the Com-


mission’s report regarding the secondary battery industry were that
competition was muted and that the two leading firms, especially
Lucas, enjoyed excessive profits, although these came mainly from
initial equipment and traction batteries, not from automotive replace-
ment batteries. The Commission thought that resale price maintenance
and the operation of the information agreement had contributed to
restricting competition in the industry, and recommended that they
be abandoned. The information agreement was indeed abandoned
shortly afterward, while resale price maintenance continued for a
few years before it was dropped as a consequence of the 1964 Resale
Prices Act.
The available evidence on prices and profits indicates that by the
mid-1960s competition had intensified in the industry and that this
trend continued in the following years. In fact, Swann et al. argue
that the market power of the leading firms was already being eroded
in the early 1960s because of competition from smaller producers
of cheap batteries. Between 1963 and 1970, the prices of a leading
manufacturer increased by only about 20% while the cost of materials
more than doubled. In addition, the large producers had to introduce
cheaper types of batteries to protect their market share against
smaller firms. According to Swann et al., firms in the industry were
able to reduce unit costs through technical improvements and better
organization in the 1960s. Still, there is little doubt that profit mar-
gins fell. Although profit data are not readily available, the report of
the National Board for Prices and Incomes on prices of secondary
batteries (NBPI 1968) states that the profitability of the larger firms
was by no means excessive in 1968, and that the increase in the cost
of materials during that year could not be absorbed without an in-
crease in prices.
While price-cost margins appear to have fallen during the 1960s,
there was no significant change in market structure. No exit of im-
portant firms and no major mergers occurred in the industry in the
1960s. According to the Census of Production, the number of do-
mestic producers of automotive batteries (which is the major product
of the industry) employing at least twenty-five persons fell from fif-
teen in 1958 to eleven in 1963, then increased to sixteen in 1968, while
the overall rate of growth of demand during this period was not
much different from the rate of growth of demand for manufacturing
as a whole. The fluctuations in firm numbers were probably driven
Price Competition and Profitability 293

by the entry and exit of smaller firms. Unfortunately, separate con-


centration data for this industry are not available. However, the con-
centration ratio for the battery industry as a whole, including primary
and secondary batteries, increased by about six percentage points
between 1958 and 1968, which is slightly less than the average for
all manufacturing (eight percentage points). This relatively modest
change is consistent with the conclusion that the 1956 Act did not
have a major effect on market structure in the secondary battery
industry.
In summary, the industry experienced an intensification of price
competition, a fall in margins (with no subsequent recovery), and no
significant change in market structure during the 1960s. What ex-
plains this evolution, which is so different from that of the trans-
former industry or the glass container industry—or of several other
industries affected by the restrictive practices legislation? The answer
is certainly not to be found in any substantial fall in endogenous
sunk costs. Both advertising and R&D were moderate in this industry,
on the order of 1%, and did not change very much during the 1960s.
Rather, the answer has to do with the initial conditions in the in-
dustry, and particularly the existing barriers to entry. In the replace-
ment battery market, the system of collective exclusive dealing seems
to have acted as a barrier to entry by smaller firms. In the initial
equipment market, the market power of the two leading firms had
been based both on their ability to offer better products and on the
existence of long-standing relationships between the battery pro-
ducers and the car manufacturers. Because of these various entry
barriers, the incumbent firms were probably earning excess profits
in the 1950s. Hence the reduction in profits brought about by the
intensification of price competition could be absorbed without the
need for a major restructuring of the industry.
It remains to be seen whether the experience of the secondary bat-
tery industry was indeed exceptional among industries affected by
the 1956 Act and therefore cannot undermine the key assumptions of
the theory set out in the introduction to this chapter. If that is the
case, it can be regarded as a counterexample that essentially confirms
the mechanism driving the theoretical predictions of this chapter,
and of the book as a whole.
This is because the evolution of the secondary battery industry
clearly illustrates the difference between a structural and a purely
behavioral interpretation of the effects of the 1956 Act on market
294 Chapter 7

structure. Recall that according to a purely behavioral interpretation,


the introduction of cartel laws would cause firms to try to restore
market power through mergers. As I argued in the concluding sec-
tion of chapter 4, this may well be true in most cases, but it cannot be
regarded as a theory of the competition-market structure relation-
ship because it does not explain how equilibrium market structure
is determined. Furthermore, the purely behavioral interpretation is
not always valid, as the evolution of the secondary battery industry
shows. On the other hand, a structural interpretation emphasizes the
fact that changes in market structure are driven by the relationship
between firms’ gross profits and sunk costs. This interpretation is
confirmed by the experience of the secondary battery industry.

7.4 The Data

As in previous chapters, the econometric analysis below will be


based on a comparison of industries affected by the 1956 Restrictive
Trade Practices Act with a control group of industries not affected.
I will use data on profits and firm/plant numbers for five different
years: 1954, 1958, 1963, 1968, and 1973. The first four of these years
are the only ones for which such data are available between the early
1950s and the late 1960s, and 1973 is the last year before the oil crisis.
As discussed in detail in previous chapters, the Act had little effect
before the first Court cases were heard in 1959, and continued to
have an effect on British industry until at least the late 1960s, and in
some cases until the early 1970s.
The industry definitions used in this chapter are sometimes at the
three-digit level of aggregation and sometimes at the four-digit level.
In particular, in most cases they are the ‘‘principal products’’ within
any three-digit ‘‘minimum list heading’’ (MLH) industry, as defined
in the individual industry reports of the UK Census of Production.
Whenever an MLH industry was not further subdivided into princi-
pal products or the subdivisions changed a lot over time, the MLH
industry was used as the industry definition for this chapter. Because
of changes over time in the Census industry or principal product
definitions, the panel is unbalanced. This section describes the con-
struction of the data set for this chapter. Once again, readers who
would prefer to skip the details on the data may well do so, since a
summary description of all the variables is given in section 7.5.
Price Competition and Profitability 295

Industry Definitions

Several of the remarks made in chapter 4 with regard to the concen-


tration data are also valid for the data used in this chapter.4 How-
ever, the data used here differ from the data used in chapters 4–6 in
one respect. The data used in chapters 4–6 were constructed on the
basis of sales of products within each industry or product group by
all plants employing at least twenty-five persons, irrespective of the
Census MLH industry to which any particular plant was allocated.
In this chapter, however, the data are constructed on the basis of
sales of plants (or firms) classified to any given principal product or
MLH industry.5 While this may be a matter of some concern in cross-
industry studies, it is far less important in studies using panel data,
such as the present one.
As mentioned above, the industry definitions used in this chapter
are sometimes MLH industries and sometimes ‘‘principal products.’’
Whenever an MLH industry was not further subdivided in the Cen-
sus reports, there was little choice but to use it as an industry defi-
nition. Most MLH industries are further subdivided into principal

4. Thus the classification of industries mainly according to supply rather than demand
characteristics is appropriate in the present context because the British cartels typically
operated within industries defined according to their supply characteristics. Also, the
fact that the data cover only UK firms is not a serious problem to the extent that the
impact of the 1956 Act was mostly on competition between UK manufacturers rather
than between manufacturing firms selling in the UK market.
5. Let me clarify this point. The UK Census of Production, like the Censuses of Pro-
duction in many other countries, allocates each plant to an MLH industry (roughly
corresponding to the three-digit level of aggregation) on the basis of the core products
of the plant. MLH industries often are further subdivided into principal products, and
each plant is allocated to a single principal product on the basis of the core product of
the plant. When computing figures for the net value of output, wages and salaries,
firm numbers, and so on, all the products produced in a certain plant are classified to
the industry that the plant is assigned to according to its core product. This means that
the reported net output for the principal product ‘‘coffee,’’ for instance, is not the actual
net output for coffee, but the total net output of all plants whose principal product is
coffee. Sales of coffee by plants classified to other industries are excluded, while sales
of products other than coffee by plants whose principal product is coffee are included.
Obviously, the profit figures, the figures for firm numbers, and so on are all subject to
the same problem.
Note that this occurs only in the data set used in the present chapter, not in any of
the data sets used in previous chapters of this book (except for the two variables that
are defined at the three-digit level in previous chapters, K/N and K/L). For instance, the
sales concentration ratio, the advertising-sales ratio, the value of sales, and so on for
coffee are derived on the basis of the total sales of coffee by all plants in manufacturing
industry.
296 Chapter 7

products, but there are several complications that may arise with
this. I will mention only the most important ones.
One complication is that the 1973 Census of Production, like all
Censuses of Production after 1968, contains no information for prin-
cipal products; all the figures are for entire MLH industries. Hence,
in some cases I had to decide whether to use the figures for a certain
MLH industry for 1954–1973 or the figures for principal products
within that MLH industry for 1954–1968. In practice, the decision to
use principal products was often imposed by the fact that several
MLH industries were ambiguous with respect to the state of compe-
tition, while some of their principal products were not. Whenever
there was a real choice, I usually preferred to use principal products,
since the gain from working with more numerous and more dis-
aggregated industry categories more than offset the loss of informa-
tion for 1973, a year not so crucial for my purposes.6
Second, the definitions of principal products often change across
years, so in several cases the figures are not comparable over the
entire period 1954–1968 (excluding minor noncomparabilities that
can be dealt with by making small adjustments). For instance, there
could be a choice between using an entire MLH as the relevant in-
dustry category, with data for all five years, and using, say, two
principal products within that MLH, with data for 1954–1963 only.
In such cases, I normally chose to use the entire MLH as my industry
definition, unless the MLH as a whole would be classified as ambig-
uous with respect to its state of competition, while some of its prin-
cipal products would not.
A third complication arises when the definition of an MLH in-
dustry changes over time. In some of these cases, it was possible to
construct comparable figures by adjusting the ones published. In
others, I could make use only of the subset of years with comparable

6. Thus, whenever the industry definition that I use is at the four-digit level (i.e., a
principal product) rather than at the three-digit level (i.e., an entire MLH industry), the
1973 observation is not available. Does this lead to any bias in the results regarding the
effect of the 1956 Act by 1973? A bias could be introduced only if MLH industries were
systematically different from principal products in some respect, for instance, in the
degree of homogeneity. But there is no evidence to support such a claim: there are
several fairly heterogeneous industries which are not subdivided into principal prod-
ucts, and hence do have a 1973 observation in my data, just as there are many homo-
geneous industries which are not subdivided into principal products. Moreover, even
if it were the case that industries with an available 1973 observation were different in
some respect from industries without a 1973 observation, this would probably affect
only the estimated time effect for 1973, not the estimated effect of the 1956 Act.
Price Competition and Profitability 297

data or use one or more principal products with comparable data


within the MLH industry.
Finally, some principal product categories have very low sales or
do not correspond to any meaningful market because they contain
mostly secondary or unspecified products. To deal with this, I merged
a few small principal products into larger ones, and I also excluded
some categories altogether from the sample.7
More generally, my principal aim was to maximize the relevant
information in the data set, and this broadly involved maximizing
the number of meaningful industry categories with comparable fig-
ures for the core period 1958–1968. As in previous chapters, some
industries were excluded, either entirely or for part of the period,
because of heavy government participation or involvement (loco-
motives, aircraft, and several industries in the steel sector).

Data Sources and Variables

Several profit measures will be used in the econometric analysis of


section 7.5: the industry gross profit divided by the number of firms,
the industry gross profit divided by the number of plants, and the
price-cost margin. All these were constructed using data on net out-
put at current net producer prices, wages and salaries, firm numbers,
and plant numbers from the industry reports of the Census of Pro-
duction (various years) and from the 1973 issues of Business Monitor,
PQ series: Quarterly Statistics. The figures are for all firms employing
at least twenty-five persons.8 Certain figures were adjusted to ensure
comparability over time. Usually this involved only minor correc-
tions, although for plant numbers more elaborate corrections had to
be made. All these adjustments were similar to those applied to the
data used for the concentration regressions (see chapter 4, section 4.4).
Industry gross profit was defined as the net value of output minus
wages and salaries, and it was deflated using either the general pro-
ducer price index, obtained from the Annual Abstract of Statistics, or
industry-specific price indices, constructed in the manner described
in chapter 4. Note that my definition of gross profit includes fixed

7. In some cases, however, industry categories described as ‘‘other’’ or ‘‘remainder’’


consist primarily of a small number of well-specified products; these were included.
8. In fact, the figures for 1963 and later years are for all firms operating plants that
employ at least twenty-five persons rather than for all firms that employ at least
twenty-five persons, but the measurement error this causes is negligible.
298 Chapter 7

costs, such as capital costs, advertising expenditure, royalties, and


R&D expenditure (other than salaries of R&D personnel). Unfortu-
nately, it also includes some variable costs, the most important of
which are ‘‘other labor costs,’’ including employers’ national insur-
ance contributions, employers’ contributions to pension funds, and
so on. Data on these are not available for most years, and the implicit
assumption was made that these costs were a constant fraction of
total variable costs within each industry over the time period consid-
ered, or that any changes in these costs were uniform across industries
and will therefore be picked up by time effects.9 More generally, it is
well known that Census-based profitability data are not very precise,
but this problem is greatly alleviated by the use of panel data.
In addition to being used to construct profit measures, the number
of firms and the number of plants were used as dependent variables
in some regressions. A limitation of using these variables as mea-
sures of market structure is that they are sensitive to the number of
small firms (or plants). It is well known that small firms often do not
produce core industry products. For instance, most engineering in-
dustries contain a large number of firms that produce only parts and
accessories. Moreover, the British cartels did not usually include all
firms in any given industry, and it was often the smaller firms that
were not cartel members. Hence the effect of the 1956 Act on many
small firms in cartelized industries may have been relatively weak.
These problems are alleviated somewhat by the fact that very small
firms (firms with fewer than twenty-five employees) are excluded
from my data. Other measures of market structure, such as concen-
tration measures, are not available for the industry categories used in
this chapter.
Market size can be measured either by the net value of output or
by sales revenue, deflated either by the general producer price index
or by an industry-specific producer price index. Perhaps net output
more accurately reflects the extent of profit opportunities in an in-
dustry, and is therefore a better proxy for market size. However, its
use in profit equations is not very appropriate because of potential
endogeneity problems. Industry-specific prices may also be endoge-

9. Some statistics on labor costs in UK manufacturing as a whole, as well as in selected


sectors for 1950–1967, were published in the October 1968 issue of Economic Trends
(pp. 25–33). They show that throughout that period changes in ‘‘other labor costs’’
mirrored almost exactly the changes in wages and salaries.
Price Competition and Profitability 299

nous, and this is likely to be a more serious problem in profit equa-


tions than elsewhere in this book. On the other hand, a drawback
of the general producer price index is that it does not control for
changes in the relative prices of materials. This discussion suggests
using sales revenue deflated by the general producer price index as a
proxy for market size in profitability regressions in order to mini-
mize any potential endogeneity problems, while at the same time
checking the extent to which the results are robust to choosing dif-
ferent proxies for market size.
The competition data were fully described in chapter 3, and the
construction of the competition variable CHANGE was described in
chapter 4. Recall that CHANGE takes the value 1 for industries with a
change in competition regime and 0 otherwise (the latter group in-
cludes three industries—salt, glazed tiles, and cement—with agree-
ments continuing throughout the period).10 An econometric analysis
of the competition effect on any endogenous variable y can then be
performed by testing whether the time effects on y after 1958 are
different for the two groups of industries in regressions that control
for other factors affecting y.
As in previous chapters, I use here the capital-labor ratio, K=L, or
the capital stock of the average plant, K=N, as proxy for the cost
of setting up a plant of minimum efficient scale (net of resale value).
I use either one of these proxies in regressions with the price-cost
margin as dependent variable. However, in regressions where the
number of firms or plants is the dependent variable or directly enters
into the definition of the dependent variable, I use only K=L, since the
use of K=N would not be appropriate.11

10. The data set also contains four industries that were affected by the nationalization
of steel in 1967 and had restrictive agreements that were abandoned after 1963. Since
the nationalization of steel made the figures on firm numbers and profitability after
1967 irrelevant, only the 1954–1963 observations for these industries were included
in my data. During this period the collusive agreements were still in place in these
industries, so I chose to classify them as industries with no change in competition re-
gime. Note that the results are barely affected if these industries are dropped from the
sample. I also treated in the same way a fifth industry, telecommunications equipment
(MLH 363), where collusion continued until 1963 and then fully broke down only in
some sections of the industry. Thus I included only the 1954–1963 observations for this
industry in my data set.
11. Recall that since all the empirical models estimated in this book include industry-
specific effects, one need not assume that K/L or K/N is an accurate measure of setup
cost. All that is required is that changes in K/L or K/N are accurate measures of the
change in setup cost, which does not seem implausible.
300 Chapter 7

Estimates of net capital stock, defined as plant and machinery, are


available from O’Mahony and Oulton (1990) at the three-digit level
of aggregation (i.e., for Census MLH industries). Recall that in pre-
vious chapters my approach was to use these three-digit figures on
the assumption that changes in K=L and K=N over time would be
roughly similar for all four-digit industries within any given three-
digit industry. In fact, in previous chapters there was no alternative
but to make this assumption, because of data limitations. In the
present chapter there is an alternative, namely, to adjust the capital
stock estimates on the basis of Census data on the fraction of invest-
ment on plant and machinery accounted for by each ‘‘principal
product’’ within any given three-digit MLH industry.
A very simple adjustment was applied: the three-digit industry
capital stock was in each case multiplied by the ratio of principal
product investment to MLH industry investment, averaged over two
years. For example, my estimate for the 1963 capital stock in the
principal product ‘‘coffee’’ was the 1963 capital stock of the MLH
industry ‘‘miscellaneous food products’’ multiplied by the ratio of
investment in coffee over total investment in the MLH industry
averaged over 1958 and 1963.12 These estimates of capital stock were
then matched with employment data or with data on plant numbers
at the same level of aggregation. The figures on investment, em-
ployment, and plant numbers were taken from the Census of Pro-
duction (various years). Again, minor corrections were made to some
of these figures to ensure comparability over time.
The adjustment applied here to the capital stock figures is admit-
tedly rough. However, it should produce reasonable approximations
of capital stock, or at least of changes in capital stock, at the four-digit
industry level. As it turns out, the regression results are largely sim-
ilar whether I use the three-digit industry estimates of K=L and K=N,
as in previous chapters, or my more refined four-digit industry esti-
mates of K=L and K=N derived using the procedure outlined above.
The only significant difference between the two sets of results is that
the use of the adjusted data results in larger coefficients and t-statistics
on the setup cost proxy itself. This suggests (1) that the adjustment is

12. For 1954, the first year in the sample, it was not possible to average relative in-
vestment over two years. Therefore the three-digit industry capital stock was multi-
plied by the ratio of principal product investment to MLH industry investment for
1954.
Price Competition and Profitability 301

in the right direction and (2) that any further refinement of the capi-
tal stock estimates would not affect the results for the variables of
interest, namely, those capturing the competition effect.
Note that an additional important implication of these compari-
sons is that any measurement error in the three-digit industry esti-
mates of K=L and K=N used in previous chapters has not significantly
affected the results for the variables of interest. The results reported
in section 7.5 are those obtained using the adjusted capital stock data.
Of course, one should be aware of the limitations of these data.
These originate not only from my own rough procedure for deriving
estimates at the four-digit industry level, but also from imperfections
in the Census investment figures and the O’Mahony and Oulton
methodology for constructing the three-digit industry capital stock
estimates. These limitations have been discussed in detail in chapter
4, section 4.4. As a result, O’Mahony and Oulton are somewhat cau-
tious about the accuracy of their estimates of the level of capital stock
in different years, and tend to rely more on their estimates of the
change in capital stock across years.
Finally, a remark on the interpretation of the capital-labor ratio in
profit equations may be in order. This variable, or the capital-output
ratio, has often been used in profitability studies to control for the
fact that the endogenous variable, namely, the price-cost margin or
the rate of return on capital, includes the gross return to capital. In
the present study, the capital-labor ratio is seen as a proxy for setup
cost. This is not a real difference, however, since the setup cost is
essentially the cost of installing capital (plant and machinery).
Union power has often been found to have a positive effect on
profitability in empirical studies of the effects of unions, and this re-
sult is consistent with theoretical models of firm-union bargaining. I
therefore include in this chapter a proxy for union power among the
regressors in some specifications. The measure used is union density,
defined as the number of unionized employees over the total number
of employees. Data on this variable for the period examined here are
available at a level of aggregation between the two-digit industry
and the three-digit industry level, and were taken from Bain and
Price (1980).
A potential objection to including a proxy for union power among
the regressors is that this variable could be endogenous. For instance,
the prospect of plant closures following an increase in the intensity
302 Chapter 7

of competition might lead to a reduction of union power, to the ex-


tent that this prospect represents a bigger threat for the welfare of
workers than for that of shareholders or managers. While this argu-
ment may be correct under many circumstances, it is less obvious
that the same objection applies to union density, which is the proxy
used here: union membership will not necessarily fall when the bar-
gaining power of the union decreases as a result of changes in the
competitive environment.13 Essentially, union density is seen here
as a variable that picks up primarily exogenous influences on union
power. In any case, because of measurement and potential endoge-
neity problems, results are reported below both with and without
union density as one of the regressors.
This chapter presents results for a basic sample, pooling exogenous
sunk cost, advertising-intensive, and R&D-intensive industries, as
well as for a subsample of exogenous sunk cost industries, defined as
industries with an average or typical advertising-sales ratio of less
than 1% and an average or typical R&D-sales ratio of less than 1%
over the relevant period. The procedure used to determine these ra-
tios was essentially the same as the one used in previous chapters;
details have been given in chapter 4. The only significant new com-
plication was the generally low advertising-sales ratios for 1954,
sometimes even for industries with high advertising intensity in later
years. Since the 1954 advertising levels are often not typical, I largely
ignored them and used mostly the advertising-sales ratios during the
period 1958–1968 (or 1958–1973 whenever relevant) to classify the
industries according to their typical advertising intensity.

7.5 Empirical Models and Results

To study the joint effect of price competition on market structure


and performance, in this section I estimate reduced-form equations
derived from the theory sketched in section 7.2. According to this
theory, market structure and profits are both endogenous and deter-
mined by the same set of exogenous variables. These include market
size, the level of setup costs or scale economies, the intensity of price
competition, union power, and various time-invariant industry-

13. For instance, it may be the case that although the monetary benefits from union
membership decrease when the union’s bargaining power in wage negotiations
declines, many workers also think that being union members could reduce their
chances of being laid off.
Price Competition and Profitability 303

specific characteristics (such as the degree of product differentiation


or the elasticity of demand).
Some of the theoretical predictions of this chapter would best be
tested using data for net profit (i.e., gross profit minus fixed costs).
Unfortunately, it is not possible to construct measures of net profit
with the data available. As pointed out above, the capital stock fig-
ures are estimates, and while the estimated proportional changes in
capital stock over time in any given industry are reasonably accurate,
the estimated levels should be treated with caution. As a result, only
measures of gross profit can be used. While this is a limitation of the
present analysis, it should be emphasized that the theory also pro-
vides several predictions regarding gross profit.
Thus a key prediction is that the industry gross profit divided by
the number of plants should not change significantly following an
intensification of price competition. This will not be true if, contrary
to what the present theory maintains, cartels generally deter entry in
long-run equilibrium. Moreover, a key insight of the present theory
is that profitability should initially decline following a rise in the in-
tensity of price competition, and then be restored, or partially restored,
through a fall in firm numbers. None of the alternative theories dis-
cussed in section 7.2 can accommodate this prediction. Finally, if it
can be shown that the long-run effect of cartel laws on the industry
price-cost margin is not significant, on average, across industries,
then total industry gross profit will not fall in the long run—under
the plausible assumption that industry output does not fall as a result
of more intense price competition. Provided that total industry sunk
costs do not rise as a result of more intense competition—which, in
a model with exogenously determined setup costs, is equivalent to
saying that the number of plants does not rise—total industry net
profit will not fall in the long run: cartel laws, in this case, definitely
will not be bad for business.

The Samples and Descriptive Statistics

The basic sample of industries for this chapter contains 201 indus-
tries and 760 observations. I excluded industries with ambiguous
state of competition in 1958 (or, in a few cases, in the 1960s and early
1970s), as well as industries without at least two available observa-
tions for the core period 1958–1968 (i.e., industries with data avail-
able only for 1954–1958 or 1968–1973). The sample of exogenous
304 Chapter 7

Table 7.1
The samples used in the econometric analysis
No. of No. of
exogenous exogenous
No. of No. of sunk cost sunk cost
indus- indus- indus- indus-
tries with tries with tries with tries with
Data available for CHANGE ¼ 1 CHANGE ¼ 0 CHANGE ¼ 1 CHANGE ¼ 0
All five years 16 30 (1 collusive) 10 15 (1 collusive)
1954, 1958, 1963, 39 49 (1 collusive) 35 32 (1 collusive)
1968
1958, 1963, 1968, 0 2 0 1
1973
1954, 1958, 1963 6 19 (3 collusive) 5 9 (2 collusive)
1958, 1963, 1968 7 1 7 0
1963, 1968, 1973 2 5 2 0
1958, 1963 4 4 (3 collusive) 3 3 (3 collusive)
1963, 1968 3 14 3 9
Total 77 124 65 69
Note: The group of industries without a change in competition regime after 1958
includes three industries where collusion continued throughout 1954–1973 and five
industries where collusion continued for the entire period for which I have data in my
sample (including four industries affected by the nationalization of steel in 1967).

sunk cost industries was derived from the basic sample by dropping
all industries with typical advertising-sales ratio (ADS) or typical
R&D-sales ratio (RDS) higher than 1% over the relevant period; it
contains 134 industries and 502 observations. One difference between
these samples and those used in previous chapters is that I have here
included all industries with available observations for only two of
the three core years 1958, 1963, and 1968—and not just those without
a change of competition regime. Table 7.1 gives details on the struc-
ture of both samples. The full set of industries and information on
key variables can be found in table B6 of appendix B.
Descriptive statistics on initial levels of profit and market structure
measures are presented in table 7.2. In particular, the table reports
means and standard deviations of five different variables in 1954 and
1958, separately for industries with a change in competition regime
after 1958 and industries without such a change. The variables are
the number of firms, NFIRMS; the number of plants, NPLANTS; the
industry gross profit (i.e., the net value of output minus wages and
salaries) divided by the number of plants and deflated by the general
Table 7.2

Price Competition and Profitability


Initial conditions (1954, 1958) in industries affected by the 1956 Act and in industries not affected
Mean (st. Mean (st. Mean (st. Mean (st. Mean (st.
deviation) of deviation) of deviation) of deviation) of deviation) of
ln NFIRMS ln NPLANTS ln PLANTPROFIT ln FIRMPROFIT PCM
All industries with CHANGE ¼ 1 ðn ¼ 55Þ
1954 4.17 (0.96) 4.53 (0.93) 4.22 (0.85) 4.57 (0.91) 0.169 (0.063)
1958 4.06 (0.92) 4.47 (0.90) 4.29 (0.89) 4.70 (0.95) 0.170 (0.061)
All industries with CHANGE ¼ 0 ðn ¼ 79Þ
1954 4.24 (1.16) 4.56 (1.08) 4.09 (1.10) 4.39 (1.23) 0.181 (0.080)
1958 4.10 (1.13) 4.43 (1.08) 4.20 (1.15) 4.51 (1.25) 0.175 (0.078)
Exogenous sunk cost industries with CHANGE ¼ 1 ðn ¼ 45Þ
1954 4.20 (0.95) 4.54 (0.93) 4.05 (0.77) 4.39 (0.83) 0.168 (0.065)
1958 4.08 (0.90) 4.48 (0.90) 4.08 (0.80) 4.48 (0.86) 0.168 (0.064)
Exogenous sunk cost industries with CHANGE ¼ 0 ðn ¼ 47Þ
1954 4.55 (1.01) 4.86 (0.94) 3.65 (0.86) 3.95 (1.00) 0.156 (0.050)
1958 4.43 (1.00) 4.74 (0.93) 3.65 (0.84) 3.98 (0.98) 0.144 (0.042)
Notes: The figures are based on industries with available data for both 1954 and 1958. The figures for ln FIRMS and ln FIRMPROFIT are based
on 77 (rather than 79) industries with CHANGE ¼ 0 and 46 (rather than 47) exogenous sunk cost industries with CHANGE ¼ 0. The group
of industries without a change in regime includes 8 collusive industries (7 in the exogenous sunk cost sample). n indicates the number of
industries.

305
306 Chapter 7

producer price index for all manufacturing, PLANTPROFIT; the de-


flated industry gross profit divided by the number of firms, FIRM-
PROFIT; and the price-cost margin, PCM, defined as the net value of
output minus wages and salaries divided by sales revenue. For the
first four of these variables a log transformation is used. The figures
are based on industries with available data for both 1954 and 1958,
and therefore can also provide information on the evolution of profits
and market structure across different groups of industries before the
implementation of the 1956 legislation.
The first thing to note is that there is little difference in initial con-
ditions between industries affected by the 1956 Act and those in the
control group, especially when one looks at the full sample. Note
that the price-cost margin is slightly higher for industries in the
control group on the basis of the full sample, but lower when focus-
ing on exogenous sunk cost industries. This is partly due to the fact
that advertising-intensive and R&D-intensive industries tend to have
higher price-cost margins and also are less likely to be collusive than
exogenous sunk cost industries.
Even more interesting is the comparison of the 1954–1958 changes
in the two groups. The evolution of ln NFIRMS and ln NPLANTS was
very similar in the two groups, especially that of ln NFIRMS. Thus,
on the basis of the raw data at least, one can argue that any differences
observed after 1958 should be attributed to the legislation rather than
to any preexisting differential trend in the two groups. This is im-
portant not only for this chapter but also for previous chapters, since
a key assumption throughout this book has been that the results re-
garding the competition effect on market structure are not subject to
any significant bias caused by the potential endogeneity of the com-
petition variable CHANGE.14

14. It is also interesting in this respect to check the 1954–1958 change of ln NFIRMS
specifically in advertising-intensive and R&D-intensive industries. In each case, there
are only a few industries with CHANGE ¼ 1, so the results should be interpreted with
some caution. Consider first the subsample of R&D-intensive industries with available
data for 1954 and 1958. The average value of ln NFIRMS decreased from 4.55 in 1954 to
4.45 in 1958 in the six industries with CHANGE ¼ 1, while it decreased from 3.47 to
3.36 in the sixteen industries with CHANGE ¼ 0. Clearly, there is almost no difference
between the two groups.
Next, consider the subsample of advertising-intensive industries with available data
for 1954 and 1958. The average value of ln NFIRMS fell from 3.84 to 3.73 in the six
industries with CHANGE ¼ 1, while it fell from 3.80 to 3.67 in the twenty-five indus-
tries with CHANGE ¼ 0. Again, there is very little difference between the two groups.
Price Competition and Profitability 307

On the other hand, the average PCM was roughly constant between
1954 and 1958 in industries with CHANGE ¼ 1 but declined some-
what in industries with CHANGE ¼ 0. This decline was not large, at
least for the full sample: about half a percentage point. Moreover, it
can be argued that if it turns out that the average PCM declined
during 1958–1963 in industries with a change in the intensity of com-
petition relative to the control group, then this must surely be the
effect of the 1956 legislation, since the trend was, if anything, the
opposite before 1958. I will return to these issues when discussing
the regression results below.
Table 7.3 presents statistics on the average change in each of
the five endogenous variables of interest over 1958–1963 and 1963–
1968. (There are considerably fewer observations for 1973 than for
1958–1968, so descriptive statistics for 1958–1973 would be less in-
formative.) In both periods, ln NFIRMS and ln NPLANTS decreased
considerably more, on average, in industries with a change in compe-
tition regime than in industries without such a change. For the period
1958–1968 as a whole, the number of firms in the former group fell by
about 45%, while it fell by only about 20% in the latter group. Also,
the number of plants fell by about 25% in the former group, com-
pared to 8% in the latter. Of course, these comparisons do not control
for changes in other variables; still, the differences between the two
groups are very large indeed.15
With respect to the profit measures, a different picture emerges.
Consider the price-cost margin, which is perhaps the ‘‘cleaner’’ of the
three measures, since it is not directly affected by the changes in
firm and plant numbers. The most interesting figures are those for
the whole sample. In both groups of industries the average PCM in-
creased over 1958–1963 and over 1963–1968. However, during 1958–
1963 the rise was larger for industries with CHANGE ¼ 0 by about

15. Another interesting comparison is between industries with a large increase in


ln NFIRMS over 1958–1968 and industries with a large decrease in ln NFIRMS over
the same period. The comparison is limited to 144 industries with available data for
both these years. Of the five industries with the largest rise in ln NFIRMS, only one
(builders’ woodwork and prefabricated timber buildings) has CHANGE ¼ 1, while
four (tufted carpets, carpeting, and carpet floor rugs; powered industrial trucks and
industrial tractors; plastic products; fiberboard packing cases) have CHANGE ¼ 0. In
contrast, of the five industries with the largest fall in ln NFIRMS, three (lime and whit-
ing; bread and flour confectionery; spun cotton yarns) are industries with CHANGE ¼ 1,
and two (loom state cloth of cotton; tiles, other than of precast concrete and brick-earth)
have CHANGE ¼ 0.
Table 7.3

308
Average change in ln NFIRMS, ln NPLANTS, ln PLANTPROFIT, ln FIRMPROFIT, and PCM, 1958–1963 and 1963–1968
D ln NFIRMS D ln NPLANTS D ln PLANTPROFIT D ln FIRMPROFIT DPCM
All industries with CHANGE ¼ 1 ðn ¼ 62Þ
1958–1963 0.18 (0.24) 0.09 (0.23) 0.35 (0.30) 0.44 (0.30) 0.024 (0.033)
1963–1968 0.20 (0.22) 0.15 (0.22) 0.29 (0.25) 0.33 (0.29) 0.012 (0.035)
All industries with CHANGE ¼ 0 ðn ¼ 82Þ
1958–1963 0.09 (0.32) 0.03 (0.34) 0.40 (0.26) 0.47 (0.28) 0.032 (0.033)
1963–1968 0.09 (0.22) 0.05 (0.22) 0.24 (0.31) 0.27 (0.31) 0.004 (0.044)
Exogenous sunk cost industries with CHANGE ¼ 1 ðn ¼ 52Þ
1958–1963 0.19 (0.25) 0.10 (0.24) 0.35 (0.29) 0.44 (0.29) 0.024 (0.030)
1963–1968 0.22 (0.22) 0.16 (0.22) 0.28 (0.26) 0.33 (0.30) 0.011 (0.036)
Exogenous sunk cost industries with CHANGE ¼ 0 ðn ¼ 48Þ
1958–1963 0.12 (0.24) 0.07 (0.26) 0.43 (0.22) 0.48 (0.24) 0.033 (0.022)
1963–1968 0.11 (0.19) 0.08 (0.19) 0.23 (0.29) 0.26 (0.25) 0.012 (0.038)
Notes: The figures are based on industries with available data for 1958, 1963, and 1968. The figures in parentheses are standard deviations.
n denotes the number of industries.

Chapter 7
Price Competition and Profitability 309

one percentage point, while the exact opposite is the case for 1963–
1968. In other words, the average PCM in industries with CHANGE
¼ 1 fell during 1958–1963 by about one percentage point relative to
the control group of industries, but it recovered during 1963–1968.16
In exogenous sunk cost industries, on the other hand, we still have
a relative fall in PCM during 1958–1963 in industries where price
competition intensified, but no obvious recovery during 1963–1968.
Admittedly, these changes are not large, and it is difficult to draw
any conclusions on the basis of these statistics alone. To unravel the
link between changes in market structure and changes in profit-
ability in industries affected by the 1956 Restrictive Trade Practices
Act, I now turn to the econometric analysis.

The Econometric Model

As in previous chapters, I will use a panel data model with indi-


vidual-specific effects. These control for industry characteristics that
affect market structure and profitability but are relatively stable over
time, such as the degree of horizontal differentiation and the elasticity
of demand. Time dummies will also be included among the regressors
in an attempt to control for other factors that may have influenced
the evolution of market structure and profitability over the period
examined, such as changes in the tax system in the mid-1960s that
are thought to have encouraged mergers; improvements in the system
of transport; economies of scale in product development, distribution,
and the raising of finance; the progressive opening of the British
economy; the UK government’s prices and incomes policies between
1965 and 1973; and macroeconomic fluctuations. It is very difficult to
measure these factors at the industry level, but it is plausible to as-
sume that their effect would have been more or less realized equally
across all industries, or at least that there would not be a systematic

16. Two of the five industries with the largest increase in PCM between 1958 and 1968
have CHANGE ¼ 1: clay, brick-earth, marl, shale, and chalk; and domestic hollow-
ware. The other three are finished thread; electrical equipment for motor vehicles,
cycles, and aircraft, excluding accumulators (secondary batteries); and powered in-
dustrial trucks and industrial tractors. Similarly, two of the five industries with the
largest decrease in PCM between 1958 and 1968 have CHANGE ¼ 1: cotton waste
yarns; and electrical ware of porcelain, earthenware, or stoneware. The other three are
British wines, alcoholic cider, and perry (fermented pear juice); photographic and
document copying equipment; and margarine.
310 Chapter 7

difference between previously collusive and noncollusive industries


with respect to these factors. Some indirect evidence supporting this
statement for the case of import competition was discussed in chap-
ter 4.
The basic specification for the number of firms is

ln NFIRMSit ¼ a i þ b1 ln SSit þ b2 lnðK=LÞit þ b3 Y54 þ b4 Y63 þ b 5 Y68


þ b6 Y73 þ b7 CHANGE  Y54 þ b8 CHANGE  Y63
þ b9 CHANGE  Y68 þ b 10 CHANGE  Y73 þ uit ;

and similarly for the number of plants. SS is the industry sales reve-
nue deflated by the general producer price index;17 K=L is the capital-
labor ratio; Y54, Y63, Y68, and Y73 are time dummies for 1954, 1963,
1968, and 1973, respectively; and the interaction terms should pick
up any differences after 1958 between industries with a change in
competition regime and industries without such a change. Thus the
coefficient on CHANGE  Y63 (CHANGE  Y68, CHANGE  Y73)
measures the effect of the 1956 Act between 1958 and 1963 (1968,
1973). The benchmark year is 1958, since it is generally accepted that
the Act had little effect on competition before then. The coefficient on
CHANGE  Y54 serves as an indirect check of the presumption that
the evolution of market structure during 1954–1958 was not signifi-
cantly different between the two groups of industries. CHANGE is
defined here (and in the profit regressions below) according to the
industry categories used for the endogenous variables. In some
regressions the variable UNION, which denotes union density, is also
included.
The estimated model for each of the three profit measures defined
above is
ln Profit measureit ¼ a i þ g1 ln SSit þ g2 lnðK=xÞit þ g3 Y54 þ g4 Y63 þ g5 Y68

þ g6 Y73 þ g7 CHANGE  Y54 þ g8 CHANGE  Y63


þ g9 CHANGE  Y68 þ g10 CHANGE  Y73 þ eit ;

where the ‘‘profit measure’’ is ln PLANTPROFIT, ln FIRMPROFIT or

17. As mentioned in section 7.4, the potential endogeneity of prices, especially in profit
regressions, suggests using the general producer price index for all manufacturing as a
deflator of sales revenue (a measure of market size). Using industry-specific price in-
dices as deflators gave results broadly similar to those reported here.
Price Competition and Profitability 311

PCM; ‘‘K/x’’ is either the capital stock of the average plant, K=N, or
the capital-labor ratio, K=L; and the other variables are the same as
above. Again, UNION is sometimes also included as an additional
regressor.18
This specification is very different from those typically used in
‘‘traditional’’ studies of the link between market structure and prof-
itability (such as Cowling and Waterson 1976, or Hart and Morgan
1977), or in the literature on the effects of economic integration on
profitability. In these studies, a measure of market structure is always
included among the regressors. My specification, on the other hand, is
a reduced-form equation derived from a theoretical model in which
market structure and profit are both endogenous. As pointed out in
the introduction to this chapter, the model’s predictions regarding
the effect of a change in the intensity of price competition on profit-
ability depend on allowing market structure to change to restore the
long-run equilibrium. It is therefore important for testing these pre-
dictions that one does not control for changes in market structure
when specifying the profit equation.
Note that a simultaneous-equations approach cannot be used here
because it is difficult to find any variable that affects the number of
firms and does not also influence profitability. Nevertheless, the
reduced-form equations can still provide important insights on the
interaction between market structure and profitability through a
comparison of short-run and long-run effects of competition, as will
be shown later in this chapter.19

18. The denominator of PCM is sales revenue. Some studies (e.g., Hart and Morgan
1977, Conyon and Machin 1991) have used net output as the denominator of PCM on
the grounds that net output, unlike sales revenue, is not influenced by input prices,
duties and subsidies, and the degree of vertical integration within an industry. These
arguments are more important for cross-section studies than for studies using panel
data. In any case, regressions using this alternative definition of PCM gave results
similar to those reported here.
19. Let me also point out again that a dynamic panel data model cannot be used in the
present context because of data limitations. Since the years in my panel are separated
by periods of four to five years, however, it is not clear that there should be any sig-
nificant effect of lagged values of the endogenous variables because of adjustment lags
or for other reasons. Also, some of the other arguments advanced in the context of the
concentration regressions of chapter 4 may be of relevance here as well, namely, that
the econometric specification used allows the competition effect to operate with a lag,
that setup costs are measured somewhat imprecisely anyway, and that there is no
evidence of significant serial correlation in the residuals of the models estimated in
this chapter.
312 Chapter 7

Estimation Results

The model has been estimated for the whole sample as well as for the
sample of exogenous sunk cost industries and the results are presented
in tables 7.4–7.6. All the results are for a fixed-effects specification.20
The reported standard errors are heteroskedasticity-consistent, ad-
justed for small sample bias following MacKinnon and White (1985).
As in previous regressions using fixed-effects estimation, two differ-
ent R 2 ’s are reported: the first does not include the fixed industry
effects, while the second does.21
Table 7.4 contains regression results for ln NFIRMS and
ln NPLANTS. Note that in the regressions using the whole sample,
interaction variables are used to control for possible differences be-
tween exogenous sunk cost, advertising-intensive, and R&D-intensive
industries with respect to the effect of market size on the number of
firms or plants. In particular, AD2  ln SS ðRD2  ln SSÞ is equal to
ln SS for industries with typical or average ADS (RDS) higher than
2% over the relevant period and to 0 for industries with ADS (RDS)
lower than 2%.22
The results in table 7.4 suggest that the 1956 Act had a strong
and statistically significant negative effect on the number of firms
in the long run. This effect was only partly realized by 1963, and it
was mostly realized by 1968. The magnitude of the coefficient on
CHANGE  Y73 implies that the intensification of price competition
following the 1956 Act reduced the number of firms by about 12%
in the average industry between 1958 and 1973. For exogenous sunk

20. The Hausman test always rejects the random-effects model. In any case, the results
from this model with respect to the competition effect are similar to those obtained
using fixed-effects estimation.
21. The reported results are for models with standard i.i.d. residuals. I also estimated a
model with an AR1 error structure, and the results were very similar to those reported
here. The estimated coefficient of serial correlation in these regressions was very low
(and often even negative): around 0.05 in market structure regressions, and between
0.05 and 0.20 in profit regressions.
22. I also experimented with alternative interaction variables—using 1% instead of 2%
as the cutoff point—but these were not statistically significant. In preliminary regres-
sions, I also included such interaction variables for the competition effect, but they
were not statistically significant, either individually or jointly. Note that whether
an industry’s typical advertising-sales ratio or R&D-sales ratio over a period of ten or
twenty years is higher or lower than 2% is largely determined by exogenous charac-
teristics such as advertising effectiveness and technological opportunity, so the inter-
action variables are not endogenous.
Price Competition and Profitability 313

cost industries in particular, the effect was even stronger: a fall of


about 15% or more. This may understate the impact of competition
to the extent that there is measurement error in the construction of
CHANGE as a result of ineffective or unregistered agreements.
The effect of competition on the number of plants was some-
what weaker: the coefficients on CHANGE  Y68 and CHANGE  Y73
in regressions with ln NPLANTS are negative but smaller in abso-
lute value than the corresponding coefficients in regressions with
ln NFIRMS, and usually are not statistically significant at the 5%
level. This implies that while much of the structural adjustment in
British manufacturing following the 1956 legislation took the form of
exit, mergers were also part of the process, and hence the reduction
in firm numbers was more pronounced than the reduction in plant
numbers. (Recall that the theory developed in this book allows for
both exit and mergers as a result of tougher competition.)
Finally, note that the coefficient on the capital-labor ratio is nega-
tive and usually statistically significant at the 5% level in these regres-
sions, while that on market size is positive and statistically significant
at the 1% level. A comparison with the coefficients on the interaction
variables AD2  ln SS and RD2  ln SS suggests that market size has a
positive effect on firm or plant numbers across all classes of indus-
tries, although this effect is less pronounced in advertising-intensive
industries.23
Table 7.5 presents results for ln PLANTPROFIT and ln FIRMPROFIT.
The first of these variables may be the one more closely associated
with the theory described in section 7.2. This is because the capital-
labor ratio is meant to control for setup costs at the plant level. (Note,
in this respect, that the coefficient on ln K=L, which is everywhere
positive and statistically significant, is larger in regressions with
ln PLANTPROFIT than in regressions with ln FIRMPROFIT.) To the
extent that the plant-to-firm ratio increases, gross profit per firm
could rise relative to gross profit per plant if there are significant
economies of multiplant operation. The results in table 7.5 provide
no evidence of any significant impact of the 1956 Act on the gross
profit of the average plant, which is consistent with the theory. More-

23. It is not surprising that differences across classes of industries with respect to
the market size effect on firm numbers are not significant. Numerous small firms in
advertising-intensive and R&D-intensive industries spend little or nothing on advertis-
ing or R&D, and may produce secondary industry products. The endogenous sunk cost
models of chapters 5 and 6 (or of Sutton 1991, 1998) are not relevant for these firms.
Table 7.4

314
Regression results for ln NFIRMS and ln NPLANTS (fixed-effects estimation)
Dependent variable: ln FIRMS Dependent variable: ln PLANTS
All industries Exogenous sunk cost industries All industries Exogenous sunk cost industries
ln SS 0.582 0.581 0.518 0.515 0.658 0.658 0.605 0.604
(0.041) (0.041) (0.041) (0.040) (0.031) (0.031) (0.031) (0.031)
ln K=L 0.065 0.064 0.108 0.110 0.125 0.124 0.138 0.138
(0.047) (0.047) (0.055) (0.055) (0.041) (0.041) (0.048) (0.048)
UNION — 0.346 — 0.514 — 0.209 — 0.194
(0.293) (0.381) (0.226) (0.286)
Y54 0.113 0.119 0.080 0.086 0.100 0.103 0.063 0.065
(0.024) (0.025) (0.030) (0.030) (0.024) (0.024) (0.028) (0.029)
Y63 0.182 0.177 0.191 0.184 0.102 0.099 0.114 0.111
(0.022) (0.022) (0.026) (0.026) (0.021) (0.022) (0.025) (0.025)
Y68 0.351 0.340 0.322 0.310 0.234 0.228 0.226 0.222
(0.034) (0.034) (0.041) (0.040) (0.032) (0.032) (0.039) (0.038)
Y73 0.450 0.408 0.401 0.355 0.381 0.356 0.319 0.302
(0.051) (0.059) (0.061) (0.067) (0.045) (0.052) (0.053) (0.056)
CHANGE  Y54 0.004 0.005 0.028 0.030 0.054 0.054 0.028 0.027
(0.040) (0.040) (0.045) (0.045) (0.035) (0.035) (0.041) (0.041)
CHANGE  Y63 0.037 0.035 0.021 0.017 0.025 0.024 0.014 0.012

Chapter 7
(0.031) (0.031) (0.035) (0.035) (0.029) (0.029) (0.034) (0.033)
CHANGE  Y68 0.095 0.086 0.110 0.090 0.063 0.057 0.062 0.054
(0.042) (0.043) (0.048) (0.052) (0.035) (0.036) (0.040) (0.043)
CHANGE  Y73 0.133 0.119 0.190 0.155 0.080 0.071 0.128 0.115

Price Competition and Profitability


(0.061) (0.062) (0.066) (0.071) (0.054) (0.055) (0.061) (0.064)
AD2  ln SS 0.305 0.297 — — 0.300 0.296 — —
(0.078) (0.078) (0.066) (0.067)
RD2  ln SS 0.087 0.095 — — 0.053 0.048 — —
(0.081) (0.082) (0.073) (0.073)
R2 0.64 0.65 0.68 0.68 0.65 0.65 0.65 0.65
R 2LSDV 0.98 0.98 0.98 0.98 0.98 0.98 0.98 0.98
Hausman statistic 159.8 162.9 143.6 142.7 127.4 130.9 74.4 70.1
Prob-value A0 A0 A0 A0 A0 A0 A0 A0
No. of industries 201 201 134 134 201 201 134 134
No. of industries 77 77 65 65 77 77 65 65
with CHANGE ¼ 1
No. of observations 758 758 501 501 758 758 501 501
Note: Heteroskedasticity-consistent standard errors in parentheses.

315
Table 7.5

316
Regression results for ln PLANTPROFIT and ln FIRMPROFIT (fixed-effects estimation)
Dependent variable: ln FIRMPROFIT Dependent variable: ln PLANTPROFIT
All industries Exogenous sunk cost industries All industries Exogenous sunk cost industries
ln SS 0.441 0.440 0.455 0.453 0.363 0.362 0.365 0.362
(0.058) (0.058) (0.059) (0.058) (0.054) (0.054) (0.051) (0.050)
ln K=L 0.146 0.147 0.154 0.152 0.200 0.202 0.182 0.179
(0.055) (0.055) (0.065) (0.065) (0.052) (0.052) (0.059) (0.060)
UNION — 0.532 — 0.446 — 0.638 — 0.767
(0.389) (0.421) (0.316) (0.361)
Y54 0.078 0.070 0.018 0.012 0.068 0.058 0.003 0.007
(0.036) (0.036) (0.039) (0.039) (0.035) (0.035) (0.038) (0.037)
Y63 0.347 0.354 0.379 0.385 0.268 0.277 0.302 0.314
(0.031) (0.031) (0.036) (0.036) (0.031) (0.030) (0.036) (0.036)
Y68 0.524 0.542 0.575 0.585 0.411 0.431 0.481 0.498
(0.046) (0.046) (0.052) (0.051) (0.044) (0.044) (0.050) (0.049)
Y73 0.616 0.680 0.694 0.734 0.566 0.641 0.616 0.685
(0.067) (0.070) (0.075) (0.077) (0.059) (0.066) (0.069) (0.074)
CHANGE  Y54 0.043 0.042 0.085 0.083 0.017 0.018 0.026 0.024
(0.057) (0.057) (0.062) (0.062) (0.054) (0.054) (0.060) (0.059)
CHANGE  Y63 0.014 0.011 0.031 0.028 0.026 0.022 0.038 0.033

Chapter 7
(0.046) (0.046) (0.050) (0.050) (0.045) (0.045) (0.050) (0.050)
CHANGE  Y68 0.069 0.082 0.028 0.045 0.035 0.052 0.020 0.009
(0.055) (0.057) (0.061) (0.065) (0.050) (0.051) (0.056) (0.059)
CHANGE  Y73 0.092 0.114 0.115 0.145 0.012 0.040 0.035 0.088

Price Competition and Profitability


(0.086) (0.090) (0.086) (0.094) (0.079) (0.081) (0.087) (0.094)
AD2  ln SS 0.250 0.261 — — 0.242 0.257 — —
(0.125) (0.125) (0.112) (0.113)
RD2  ln SS 0.232 0.220 — — 0.096 0.081 — —
(0.114) (0.115) (0.110) (0.112)
R2 0.78 0.78 0.80 0.80 0.74 0.75 0.74 0.74
R 2LSDV 0.97 0.97 0.97 0.97 0.97 0.97 0.96 0.96
Hausman statistic 70.3 79.7 64.1 65.5 46.8 60.5 39.1 45.0
Prob-value A0 A0 A0 A0 A0 A0 A0 A0
No. of industries 201 201 134 134 201 201 134 134
No. of industries 77 77 65 65 77 77 65 65
with CHANGE ¼ 1
No. of observations 758 758 501 501 758 758 501 501
Note: Heteroskedasticity-consistent standard errors in parentheses.

317
318 Chapter 7

over, there is no evidence of any significant impact of the Act on firm


gross profit either.
An interesting feature of the results in tables 7.4 and 7.5 is the
magnitude of the coefficients on the year dummies. After controlling
for market size, the capital-labor ratio, union density, the effect of
cartel policy, and industry effects, the number of plants or firms in any
given industry in 1973 was, on average, about 30–40% lower than in
1958, while the gross profit of the average plant or firm was more
than 50% higher. This evolution seems to have continued an existing
trend during 1954–1958.
To some extent the high coefficients on the year dummies must be
due to the crudeness of the setup cost proxy used here. Since K=L
was increasing across industries throughout the period, it is corre-
lated with the time dummies; thus, to the extent that K=L is only an
imperfect proxy for setup costs, and these were also increasing, their
effect could be partly picked up by the time dummies. Note, in this
respect, that K=L is not statistically significant at the 5% level in some
regressions for ln FIRMS, and that in random-effects specifications
its explanatory power is considerably higher while that of the time
dummies is somewhat lower. Moreover, the time dummies may be
capturing the effect of scale economies not directly associated with
the cost of plant and machinery, such as scale economies in market-
ing or the raising of finance. Finally, some of the apparent explanatory
power of the time dummies may simply be due to the unbalanced
structure of the panel.
Still, it is difficult to escape the conclusion that much of the fall
in firm and plant numbers during this period was due to factors not
explicitly included in the present theory. This, of course, does not
invalidate the comparison between industries affected by the 1956
Act and industries not affected, to the extent that these other factors
are not correlated with the variable CHANGE.
While the results in table 7.5 are consistent with the theory, the
discussion in the previous paragraph and the fact that the overall
changes in ln PLANTPROFIT and ln FIRMPROFIT reflect to a large
extent the significant decrease in firm and plant numbers during the
1950s and the 1960s suggest that it is also necessary to consider other
profit measures in order to assess the impact of the 1956 Act on firms’
profits. A measure not directly influenced by firm or plant numbers,
such as the price-cost margin, should be particularly useful.
Price Competition and Profitability 319

Table 7.6 reports the results. The first thing to note is that the
coefficients on CHANGE  Y68 and CHANGE  Y73 are small and
nowhere statistically significant, even at the 10% level. The failure to
detect any long-run effect of price competition on the price-cost
margin is consistent with the theory developed in this book. In par-
ticular, it justifies the Selten-Sutton emphasis on the effect of firm
conduct on market structure rather than on profits. Of the other
variables, ln K=L and ln K=N have a positive effect on PCM, while
union density has a negative effect, as expected. The time dummies
may again be picking up some of the effect of scale economies on the
price-cost margin, given that ln K=L and ln K=N are imperfect proxies
for setup cost. These results are hardly affected if the sales variable,
which is nowhere statistically significant, is dropped.
It is also very interesting to compare the short-run and the long-
run impact of competition. This may be the most decisive test of the
present theory, which predicts a particular link between the evolu-
tion of market structure and the evolution of profitability in the short
run and in the long run. Table 7.4 shows that the effect of the 1956
Act on market structure was modest over the period 1958–1963;
on the other hand, it was between 1963 and 1968 that most of the
restructuring of previously cartelized industries occurred. Moreover,
the overall picture from table 7.6 (despite small differences across
regressions) is that price-cost margins declined, on average, between
1958 and 1963 in these industries, before recovering mostly during
1963–1968. Note that the coefficient on CHANGE  Y63 is everywhere
negative and typically statistically significant at the 5% or the 10%
level, while the coefficients on CHANGE  Y68 and CHANGE  Y73
are sometimes positive, sometimes negative, and nowhere statisti-
cally significant.
This is precisely the sort of link the theory predicts between the
evolution of market structure and the evolution of profitability in
previously collusive industries: a moderate effect on the number of
firms by 1963, at which date several industries were in short-run
disequilibrium with reduced margins (including industries where
the adjustment of concentration to its long-run value was being
delayed by a slow rate of depreciation of the capital stock); then a
significant negative effect on firm numbers between 1963 and 1968,
leading to a rise in price-cost margins in these industries.
The coefficient on CHANGE  Y63 in regressions with PCM, al-
though typically significant at the 5% or 10% level, is nevertheless
Table 7.6

320
Regression results for PCM (fixed-effects estimation)
Dependent variable: PCM
All industries Exogenous sunk cost industries
ln SS 0.008 0.006 0.008 0.005 0.004 0.002 0.003 0.001
(0.007) (0.006) (0.006) (0.006) (0.007) (0.007) (0.006) (0.006)
ln K=L 0.020 — 0.021 — 0.014 — 0.014 —
(0.007) (0.007) (0.008) (0.008)
ln K=N — 0.012 — 0.013 — 0.007 — 0.007
(0.006) (0.006) (0.007) (0.007)
UNION — — 0.177 0.176 — — 0.186 0.188
(0.042) (0.041) (0.051) (0.051)
Y54 0.006 0.005 0.008 0.008 0.010 0.009 0.012 0.011
(0.004) (0.004) (0.004) (0.004) (0.005) (0.005) (0.005) (0.005)
Y63 0.029 0.030 0.031 0.033 0.030 0.032 0.033 0.034
(0.004) (0.004) (0.004) (0.004) (0.005) (0.005) (0.005) (0.005)
Y68 0.026 0.030 0.032 0.035 0.036 0.040 0.040 0.043
(0.006) (0.006) (0.005) (0.005) (0.006) (0.006) (0.006) (0.006)
Y73 0.021 0.030 0.042 0.050 0.040 0.046 0.057 0.062
(0.009) (0.009) (0.009) (0.009) (0.011) (0.010) (0.010) (0.009)
CHANGE  Y54 0.003 0.003 0.003 0.003 0.005 0.005 0.005 0.005

Chapter 7
(0.007) (0.007) (0.006) (0.006) (0.007) (0.007) (0.007) (0.007)
CHANGE  Y63 0.012 0.012 0.011 0.011 0.011 0.011 0.009 0.009
(0.006) (0.006) (0.006) (0.006) (0.007) (0.007) (0.007) (0.007)
CHANGE  Y68 0.004 0.004 0.001 0.001 0.010 0.010 0.003 0.003

Price Competition and Profitability


(0.006) (0.006) (0.007) (0.006) (0.007) (0.007) (0.008) (0.008)
CHANGE  Y73 0.003 0.011 0.004 0.003 0.008 0.014 0.004 0.001
(0.011) (0.011) (0.011) (0.011) (0.013) (0.013) (0.013) (0.014)
AD2  ln SS 0.013 0.015 0.009 0.012 — — — —
(0.015) (0.015) (0.015) (0.015)
RD2  ln SS 0.025 0.026 0.021 0.022 — — — —
(0.014) (0.014) (0.014) (0.014)
R2 0.28 0.28 0.31 0.30 0.33 0.32 0.36 0.35
R 2LSDV 0.89 0.89 0.89 0.89 0.88 0.88 0.88 0.88
Hausman statistic 29.6 29.2 39.6 37.5 19.8 17.7 28.9 26.6
Prob-value 0.003 0.004 0.0002 0.0004 0.03 0.06 0.002 0.005
No. of industries 201 201 201 201 134 134 134 134
No. of industries 77 77 77 77 65 65 65 65
with CHANGE ¼ 1
No. of observations 760 758 760 758 502 501 502 501
Note: Heteroskedasticity-consistent standard errors in parentheses.

321
322 Chapter 7

not large: about one percentage point. This is consistent with the
evidence from the descriptive statistics of table 7.3. It has to be borne
in mind that only a subset of the previously cartelized industries
were in short-run disequilibrium in 1963. In several industries com-
petition had not yet emerged, and there must have been several
others where competition had emerged and much of the adjustment
of market structure had already taken place. Thus the magnitude of
the coefficient on CHANGE  Y63 should not be taken as a measure of
the fall in the price-cost margin following a change of competition
regime and prior to any adjustment of market structure; this is likely
to be much larger than one percentage point.
Several additional remarks are in order. First, the econometric
analysis confirms that the potential endogeneity of CHANGE does
not seem to be a source of significant bias in the results. Recall that
an indirect check of this claim is to compare the evolution of market
structure and profitability between industries with CHANGE ¼ 1
and industries with CHANGE ¼ 0 before 1958. I have already empha-
sized that the descriptive statistics provide no evidence that the dif-
ferences between the two groups after 1958 could be attributed to a
preexisting differential trend. This is confirmed by the econometric
analysis. Thus the coefficient on CHANGE  Y54 is nowhere statisti-
cally significant, even at the 20% level, suggesting that there was no
difference in the evolution of market structure and profitability be-
tween the two groups of industries before 1958 (the benchmark year).
Moreover, the coefficient on CHANGE  Y63 usually has the same sign
as the coefficient on CHANGE  Y54, which implies that even if a
differential trend between the two groups of industries had existed
before 1958 for some variables and is imprecisely measured (because
the 1954–1958 period is too short, say), this trend was, if anything,
reversed during 1958–1963.
Second, a comparison of the R 2 ’s in table 7.4 with those in table 7.6
reveals an interesting feature of the regressions presented in this
chapter. In both these tables two R 2 ’s are reported and, as pointed
out previously, the difference between the two is a measure of the
explanatory power of the industry effects. It can be seen that when
the R 2 is defined to exclude fixed industry effects, it is very much
higher in regressions with ln NFIRMS and ln NPLANTS than in re-
gressions with PCM. On the other hand, when the R 2 is defined to
include fixed industry effects, it is only slightly higher in regressions
Price Competition and Profitability 323

with ln NFIRMS and ln NPLANTS than in regressions with PCM.


In other words, the industry-specific effects have far more explana-
tory power for the price-cost margin than for firm or plant numbers.
The importance of time-invariant industry-specific characteristics in
regressions with PCM is consistent with existing evidence according
to which the variation in price-cost margins over time within the
typical industry is modest when compared to the between-industry
variation (see Geroski 1991). The interesting question therefore is
why this is not also the case for firm and plant numbers in the pres-
ent context. One explanation may be the significant decline in firm
and plant numbers throughout the period examined here; this de-
cline is largely picked up by the time dummies and thus raises the R 2
(excluding the industry effects) in regressions with ln NFIRMS and
ln NPLANTS.
Finally, it could be argued that the profit equations estimated above
do not adequately control for industry-specific factors that may cause
departures from long-run equilibrium. Hence I also ran regressions
including among the regressors the variable D ln SS, defined for in-
dustry i and year t as the change in ln SS in the five-year period pre-
ceding year t. A disadvantage of this alternative specification was
that the first-year observation for each industry could not be used,
and this implied dropping all 1954 observations. In any case, the co-
efficient on D ln SS was everywhere positive and sometimes statisti-
cally significant, but the rest of the results did not change.

7.6 Concluding Remarks

This chapter has examined the joint evolution of market structure


variables, such as the number of firms and the number of plants, and
profit measures, such as the average firm profit, the average plant
profit, and the profit margin, in UK industries over the period 1954–
1973. The results support the hypothesis of a negative long-run effect
of the 1956 Act on the number of firms and of no significant long-run
effect on profitability. The former result is consistent with the finding
of a positive overall impact of the Act on concentration in previous
chapters. Moreover, the comparison of short-run and long-run effects
of competition suggests a link between changes in market structure
and profitability: in the short run, when the number of firms has not
yet fallen very much, profit margins decline; but they recover in the
long run, once the number of firms falls.
324 Chapter 7

These results are consistent with the theoretical framework devel-


oped in this book, which has emphasized the effect of price com-
petition on market structure and nonprice competition rather than on
profitability. The results say that in long-run equilibrium, most car-
tels will result in excess entry rather than excess profits relative to the
absence of collusion. Legislation prohibiting cartels will therefore
reduce the number of firms rather than their profits.24
The results also imply that most cartels cannot restrict entry in the
long run—or at least they cannot restrict entry to a larger extent than
noncolluding firms. There are exceptions to this, as suggested by one
of the case studies described in this chapter. When the free-entry
condition is violated before the introduction of cartel policy, the
breakdown of collusion may lead to a fall in profitability and no
significant change in market structure. In this case, a reduction in
profits can be absorbed without the need for a major restructuring of
the industry, because firms were previously making excess profits. In
one sense, this case study illustrates the limitations of the present
theory, since it suggests that there can be instances where one of the
key assumptions of the theory is violated. But at the same time, it
confirms the operation of the structural mechanism driving the the-
oretical predictions and the econometric results of much of this book,
namely, the fact that changes in market structure are driven by the
relationship between firms’ gross profits and sunk costs.
One thing that has not been much emphasized here is the fact that
price-cost margins can change for two different reasons: either be-
cause prices change or because unit costs change. Throughout this
chapter I have tended to focus on changes in margins as being driven
by exogenous factors and to play down the distinction between price
changes and cost changes, on the assumption that cost changes are
largely passed on to prices, everything else being equal. Neverthe-
less, I have implicitly attributed the short-run fall in profitability in
previously collusive industries to a fall in prices caused by the abo-
lition of price-fixing agreements. On the other hand, the recovery of

24. Another interesting implication of the results presented in this chapter is that they
confirm that explicit collusion was generally not replaced by tacit collusion in the long
run in the large majority of the previously collusive industries. Recall that this is also
what the case-study evidence suggests, as pointed out in chapter 2. If explicit collusion
had been replaced by tacit collusion, then the profitability of industries affected by the
1956 Act would have increased in the long run relative to the profitability of industries
not affected by the legislation, given that the fall in firm numbers was more pro-
nounced in the former group than in the latter. But this is not what we observe.
Price Competition and Profitability 325

profitability in the longer term could be due to a recovery in prices or


a fall in unit costs or both. Again, I have implicitly attributed much
of this recovery to price increases following the drastic fall in the
number of firms. But my results are not inconsistent with the view
that margins also increased because less efficient firms could not
survive in the more competitive conditions of the 1960s, and so the
unit cost of the average firm fell. In fact, I have hinted in section 7.2
that such an effect could be part of the story, and in at least one of the
case studies discussed in section 7.3 it probably was.
At the same time, it is clear that the fall in firm numbers was a key
aspect of the evolution of industries affected by the 1956 Act. In
particular, the results of this chapter are not consistent with a strong
version of the efficiency differences approach, according to which the
main effect of more intense price competition would be the expan-
sion of low-cost firms at the expense of high-cost firms (which would
also explain why average industry profitability might not fall in the
long run) rather than a fall in the total number of firms. The results of
this chapter suggest that although the expansion of efficient firms
and a fall in unit costs may be part of the explanation for the long-
run recovery of profitability in previously collusive industries, this
recovery was largely the result of the significant restructuring of
these industries through exit and mergers.
It is also worth emphasizing that this chapter does not provide any
direct test of the free-entry zero-profit condition. What has been tested
here is not whether net profit is approximately zero in the long run
for the marginal firm; this is a very difficult task without firm-level
data on profits and capital costs. What this chapter has tested is
whether a change in firm conduct has any effect on profit in the short
run and in the long run. Thus the results presented here would also
be consistent with a model that predicts that the net profit of the
marginal firm is consistently larger than zero by a nontrivial amount,
and that this amount is not much affected by a change in conduct.
One mechanism that could account for the existence of supranormal
profits for the marginal firm is, of course, strategic entry deterrence.
However, it is not clear whether this can be assumed to work across
all industries and irrespective of whether incumbent firms collude to
fix prices or not. The results presented here seem, therefore, to sug-
gest that the zero-profit condition is a valid assumption, at least as a
first approximation, for most industries characterized by free entry.
326 Chapter 7

Still, as pointed out by Scherer and Ross (1990), this approximation


may be indeed rough—especially for R&D-intensive industries,
where firms’ capabilities often change at a rate slower than the rate
at which technology and demand conditions shift. In such industries,
and perhaps also in others, it may be questionable whether a zero-
profit equilibrium is actually achieved at any point in time. More-
over, the presence of asymmetries between firms implies that only
the marginal firm makes zero profit under free entry. More efficient
firms or firms enjoying first-mover advantages may well earn higher
than normal profits in the long run. It is, however, probably safe to
suggest, on the basis of the results presented in this book, that in the
large majority of industries, the level of excess profits in the long run
does not depend on firms’ pricing conduct, because of forces such as
entry and exit that push industries toward the zero-profit equilib-
rium. It is mainly in this sense, I think, that the free-entry zero-profit
condition is a useful approximation for the study of competition and
the determinants of market structure and profitability.
Epilogue

The search for general theories and cross-industry empirical regu-


larities has recently reemerged as a key priority in industrial eco-
nomics. It has come as a response to the rather pessimistic view that
‘‘in oligopoly, anything can happen.’’ It is, of course, true that the
results of game-theoretic models of oligopoly often depend on specific
assumptions made (including assumptions about unobservables). It
is also true that the empirical outcomes we observe are partly driven
by particular characteristics of industries (including some which are
the product of history or chance). However, the richness of theoreti-
cal results and the corresponding variety of empirical outcomes do
not imply that ‘‘anything can happen.’’ In particular, it is still possi-
ble to identify a number of mechanisms, involving observable vari-
ables, which operate in a systematic way across industries or for a
broad class of industries. This book has set out to analyze some of
these mechanisms. Accordingly, the theoretical predictions have been
tested using cross-industry data, and a number of cross-industry em-
pirical regularities have emerged.

A Brief Review of the Theory

The most fundamental mechanism analyzed in this book relates to


the joint effect of an increase in the intensity of price competition on
market structure and profitability in exogenous sunk cost industries.
Starting from a free-entry equilibrium, an intensification of price
competition—brought about by the introduction of cartel laws, eco-
nomic integration, or some other exogenous institutional change—
will cause profitability to fall, given the initial number of firms. As a
result, firms (or less efficient firms) will no longer be able to cover
their sunk costs. This will lead to a fall in firm numbers and a rise in
328 Epilogue

concentration until profitability is restored to a level that allows the


least efficient firm to cover its sunk costs. This is a clear and strong
theoretical result, and one that can be tested against the empirical
evidence using econometric analysis as well as industry case studies.
On the other hand, the theory offers no general predictions re-
garding the effect of price competition on advertising intensity, R&D
intensity, and innovative output. Specific results in this area can be
obtained only by imposing a considerable amount of structure on
theoretical models. In empirical work, one can nevertheless still ex-
amine whether any empirical regularities emerge despite the incon-
clusiveness of the theory.
What of the competition-market structure relationship in
advertising-intensive and R&D-intensive industries? Some general
results can be derived, although the theory now places only weak
constraints on the space of possible outcomes. The most likely out-
come will still be a positive effect of price competition on concentra-
tion, because a negative effect on concentration can occur only in the
event of a substantial decline in advertising or R&D intensity. Simi-
larly, profitability will most likely decline initially, following an
increase in the intensity of price competition, before it is restored
through a fall in firm numbers. Given these theoretical results, the
empirical evidence can be used for two different, but complemen-
tary, purposes. First, to confirm that the outcomes we observe on
average across industries are indeed the ones which are predicted as
being most likely by the theory; this requires the use of econometric
analysis. Second, to verify that the outcomes which are predicted as
being less likely can still be observed in certain special cases; this
requires the use of case studies.

A Brief Review of the Empirical Evidence

The introduction of cartel policy in the UK in the late 1950s was a


very rare natural experiment in industrial economics. It has given us
the opportunity to analyze the effects of competition in a systematic
way without many of the problems that typically arise when we try
to define an empirical proxy for the intensity of competition or to un-
ravel the complex interactions between competition, market struc-
ture, and performance. In particular, in this book the analysis of the
effects of competition following the introduction of cartel policy in
the UK was based on a comparison of the evolution of two clearly
Epilogue 329

defined groups of industries: those that experienced a change in


competition regime as a result of the legislation and those that were
not affected by the legislation.
Previous studies of the effects of the introduction of restrictive
practices legislation in the UK include Swann et al. (1973, 1974), Elliott
and Gribbin (1977), and O’Brien et al. (1979). Swann et al. (1973, 1974)
conducted a series of industry case studies, focusing on the effects of
the legislation on competition rather than on its repercussions on
market structure and industry performance. Their analysis is a rich
source of information on the emergence of price competition in
previously collusive British industries during the 1960s. Elliott and
Gribbin (1977) provided the first statistical results indicating a sig-
nificant effect of UK cartel policy on concentration across a wide
range of industries. Their results have been confirmed by the present
study, which has also extended their analysis in several important
ways. On the other hand, O’Brien et al. (1979), who performed a
detailed statistical analysis using firm-level data for a modest sample
of industries, found no evidence of any significant effect of the legis-
lation either on merger activity or on profitability.
The econometric results reported in this book suggest that the 1956
Restrictive Trade Practices Act, which led to the abolition of restric-
tive practices across a wide range of industries, was one of the main
factors behind the significant increase in concentration in British
manufacturing during the 1960s. In particular, the intensification of
price competition following the termination of price-fixing agree-
ments caused a rise in the five-firm concentration ratio of at least
six to seven percentage points, on average, in industries affected by
the legislation. The impact was strong in all classes of industries. A
comparison of short-run and long-run effects provides further insight
on these events. The legislation had a negative effect on profitability
of previously collusive industries during the first few years of its
implementation, when market structure had not yet fully adjusted.
However, the subsequent restructuring of these industries was asso-
ciated with a recovery of profit margins, so that profitability did not
change significantly in the long run. Thus the evidence is consistent
with the theory.
The econometric evidence also suggests that the intensification of
price competition following the 1956 Act caused, on the whole, a fall
in advertising intensity in advertising-intensive industries affected
by the legislation. Moreover, case-study evidence confirms that a fall
330 Epilogue

in concentration, following an intensification of price competition,


cannot be ruled out in high-advertising industries. However, tougher
price competition must lead to a significant fall in advertising inten-
sity if concentration does not rise or, conversely, it must cause con-
centration to rise if advertising falls little or not at all, which is
consistent with the theoretical predictions as well as with the overall
picture given by the econometric results. The negative effect of price
competition on advertising can be contrasted with the absence of any
overall effect on the number of innovations produced by firms. In
particular, there is no evidence of any significant effect of the 1956
Act on firms’ innovative output in R&D-intensive industries.

Wider Issues

Some of the wider normative issues regarding competition, such


as the effect of different competition regimes on social welfare or the
question of how much competition is socially desirable, have not
been discussed in this book because they are too difficult to examine
empirically. For instance, data limitations have precluded any anal-
ysis of the effect of the 1956 cartel legislation on prices. Moreover,
no attempt has been made to examine the effect of the 1956 Act on
the productive efficiency of firms, since this raises some conceptual
and practical issues that are difficult to resolve with the information
available. The case-study evidence reported in Swann et al. (1973,
1974), although consistent with the hypothesis of a positive effect of
the Act on the efficiency of firms, is rather sketchy. Thus the question
of the link between competition and efficiency following the break-
down of cartels in Britain remains open.
Other positive and normative implications of the present study
are, however, easier to draw. One implication is that in long-run
equilibrium most cartels will result in excess entry rather than excess
profits (relative to the absence of collusion), and hence cartel laws
will reduce the number of firms rather than their profits. A second
implication is that the recent trend toward more competition in
national and international markets is likely to increase concentration
at both the national and the international level. However, a qualifi-
cation is needed here for policy interventions that involve a reduc-
tion or removal of entry barriers in addition to an increase in the
intensity of competition: the effect of such policies on market struc-
ture will be ambiguous. A third implication is that competition is
Epilogue 331

probably not a substitute for policies that promote innovation, since


the rate of innovation does not appear to be much affected by the
degree of price competition.
What about competition policy? If high concentration and mergers
are an inevitable consequence of tougher price competition, as the
present study suggests, then there are constraints on the exercise of
merger and antitrust policies, since these policies cannot be used to
maintain or impose a market structure that is not sustainable in the
long run. Of course, to the extent that market structure is also influ-
enced by nonstructural factors, competition could lead to excessive
restructuring, so merger policy has still an important role to play.
The same is true for competition policy in general, since there is
always a real danger that dominant firms in concentrated markets
may try to abuse their market power. This brings us to the final im-
plication of the research reported in this book: the link between
competition and concentration is complex, and the present study
strengthens the case for a competition policy that focuses more on
the monitoring of conduct than on the regulation of market structure.
Indeed, a central message of this book is that when free entry is
maintained, high concentration need not be associated with high
profit margins and increased allocative inefficiency, because it may
itself be the result of more intense competition.
This page intentionally left blank
Appendix A: A Survey of
Collusive Agreements in
British Manufacturing
Industries

This appendix contains a survey of explicit collusive agreements across


British manufacturing industries from the early 1950s to the mid-1970s. It
reports all the relevant information that I have used to construct the compe-
tition data for this book. This information comes from various sources, the
most important of which have been described in chapter 3, section 3.2. It will
be presented here industry by industry, using the ‘‘minimum list heading’’
(MLH) categories of the UK 1968 Standard Industrial Classification as basic
industry categories.* I emphasize that this classification is used here merely
as a way of organizing the presentation of the material. The information on
restrictive agreements will be reported for products and subproducts within
any given MLH category.
The survey will cover only important national and regional collusive
agreements. That is, I will not be concerned with local agreements or with
agreements containing only trivial restrictions or involving a few producers
of some very specific product. Moreover, the survey will cover only agree-
ments between manufacturers. Although there were numerous important
restrictive agreements in agriculture and in distribution, and many more
between firms in the construction industry, these are not relevant to the
present study. On the other hand, the services industries, as pointed out in
chapter 2, did not fall within the scope of the 1956 Act.
When describing particular agreements, I will focus on the most important
restrictions of each. Thus, whenever there were agreed or recommended
prices or agreed quotas, I will usually not give details about ancillary restric-
tions, such as conditions of sale, quantity discounts, approved lists of dis-
tributors, or resale price maintenance. The large majority of price agreements
also specified conditions of sale, and I will not be mentioning this in each
particular case. Moreover, while a few agreements provided for net producer
prices, the large majority specified minimum or fixed ‘‘list prices’’ and com-
plemented these by a set of maximum or fixed discounts to distributors
(‘‘trade discounts’’) and other buyers. I will not be mentioning the provision
regarding trade discounts as a separate restriction in each case. Finally, note
that neither the distinction between ‘‘agreed’’ and ‘‘recommended’’ prices nor

* A glossary of British terms appears at the end of the appendixes.


334 Appendix A

the distinction between ‘‘minimum’’ and ‘‘common’’ prices carries too much
weight—especially in light of the fact that many agreements were redrafted
prior to registration. I therefore will often simply use a term such as ‘‘pricing
agreement’’ or ‘‘price-fixing agreement’’ or an equivalent expression in what
follows. On the other hand, I will always specifically mention any important
nonprice restrictions, such as market sharing, collective exclusive dealing,
aggregated rebates, and restrictions relating to advertising or R&D.
Whenever a particular agreement did not provide for price-fixing or mar-
ket sharing, I will elaborate on the types of restrictions that were in force. I
will not, however, discuss cases of individual resale price maintenance by
firms in an industry, either before or after 1956. These do not constitute
agreements under the 1956 Act; more important, as I explained in chapter 3,
there is no evidence that resale price maintenance had a significant effect on
competition between manufacturers in the absence of any other restrictions.
The question of whether a particular agreement covered the export market
in addition to the home market is often not an easy one, so it will not be
given much attention here, except for very few cases where the relevant in-
formation is both available and important for classifying an industry as col-
lusive, competitive, or ambiguous in the 1950s. Another difficult question
concerns the participation of British firms and associations in international
cartels. Since this issue is not directly relevant to the present study, it will be,
for the most part, left aside. I will report only significant cases of inter-
national cartels that are mentioned in the Register of Restrictive Trading
Agreements or the reports of the Monopolies and Restrictive Practices Com-
mission. On the other hand, I will report all the available evidence on infor-
mation agreements—although the lack of systematic data on information
agreements poses some serious problems in this respect. Thus my review of
information agreements, based on evidence contained in the register or from
case studies, will inevitably be incomplete.
The available information on significant restrictions other than pricing
restrictions is also less than complete. The main reason is that several agree-
ments were modified before registration, and this sometimes implied remov-
ing provisions regarding market sharing, collective exclusive dealing, or
aggregated rebates. This difficulty should not be overemphasized, however,
since information on such schemes is provided in several data sources other
than the register, including the 1955 Monopolies and Restrictive Practices
Commission report on ‘‘collective discrimination.’’ This report covered the
large majority of the most important instances of collective exclusive dealing,
aggregated rebates, barriers to entry into distribution, and collective resale
price maintenance—although it did not usually specify the exact restrictions
that applied to each industry listed, provided little information on coverage,
and sometimes used rather broad industry definitions.1

1. There are four such broad industry definitions in the report: ‘‘certain chemicals,’’
‘‘certain electrical products,’’ ‘‘groceries,’’ and ‘‘hardware’’; the last two should be taken
to mean ‘‘certain groceries’’ and ‘‘certain hardware products.’’
Survey of Collusive Agreements in British Industries 335

The following abbreviations will be used for some of the data sources.
Rxxx denotes the number (xxx) of an agreement in the Register of Restrictive
Trading Agreements. BT denotes the Board of Trade annual reports on the
operation of the 1948 Monopolies and Restrictive Practices Act, covering the
years 1949 to 1956. PEP denotes the Political and Economic Planning survey
of industrial trade associations, including the unpublished background ma-
terial for this survey. MRPC stands for Monopolies and Restrictive Practices
Commission, MC stands for Monopolies Commission, and MMC stands for
Monopolies and Mergers Commission (all of which are successive names
for the same public body). ‘‘Lloyds’ report’’ refers to the 1949 report of the
Lloyds’ Committee on resale price maintenance (Board of Trade 1949), and
CD refers to the 1955 MRPC report on collective discrimination. Finally, the
reports of the Restrictive Practices Court, which have been published as
Reports on Restrictive Practices Cases by the Incorporated Council of Law
Reporting, are conventionally cited as LR [volume no.] RP [page no.].
For each product, I will start with the information contained in the Register
of Restrictive Trading Agreements and then report any relevant additional
information from other sources. To avoid repetition, references to other data
sources will be made only when these sources contain additional informa-
tion relative to the register. This is typically the case with the reports of the
Monopolies and Restrictive Practices Commission, the reports of the Re-
strictive Practices Court, the 1955 MRPC report on collective discrimination,
and the case studies contained in Swann et al. (1973). It is less often the case
with the lists of collusive or allegedly collusive industries in PEP and BT.
Hence references to PEP or BT will be made below only (1) for industries
which did not register any significant restrictive agreements but are listed as
collusive or allegedly collusive in these sources, and (2) for industries whose
registered agreements do not seem significant and the detailed evidence
contained in PEP confirms that they should be treated as competitive. The BT
lists distinguish several types of restrictive practices, including price-fixing,
market sharing, and collective discrimination. Unless otherwise stated, ‘‘col-
lusion’’ should be taken to mean price-fixing when a reference is made to the
BT lists below.
I will also indicate which agreements were referred to the Restrictive
Practices Court. In particular, agreements referred to the Court will be
marked with an * next to their number on the register (i.e., Rxxx*). For those
that were struck down by the Court without being defended by the parties,
no other information will be given.2 For those that were contested by the
firms involved, I will indicate what the outcome was in each case. Brief
descriptions of all the agreements that were referred to the Court since the
introduction of the 1956 Act and summary accounts of Court proceedings for

2. I will not distinguish between agreements abandoned after being referred to the
Court (which was the usual case) and agreements referred to the Court after being
abandoned (which was a precautionary step sometimes taken by the Registrar in order
to reduce the likelihood of any future collusion by the firms involved). This distinction
is usually difficult to make on the basis of the available data.
336 Appendix A

contested cases can be found in the reports of the Registrar of Restrictive


Trading Agreements (RRTA 1961–1973) or, after 1973, in the annual reports
of the Director General of Fair Trading.
In addition to presenting the ‘‘raw’’ data on collusion, I will provide sum-
mary assessments of the effect of the 1956 Act on competition across indus-
tries and product groups. This should help clarify the link between the ‘‘raw’’
competition data and my classification of industries according to their com-
petitive status. An industry in my data sets is classified as collusive in the
1950s (or in later years) if the products subject to significant restrictions
accounted for more than 50% of total industry sales revenue. It is classified
as competitive if the products subject to significant or uncertain restrictions
accounted for less than 10% of industry sales revenue. And it is classified as
ambiguous (or the state of competition is classified as ambiguous) in all
remaining cases, namely, when the restrictions cover most of the industry
but their effect on competition is uncertain, or the data source is not fully
reliable, or the collusive products cover more than 10% but less than 50% of
total industry sales, and so on.
Furthermore, an industry in my data sets is defined as one with a change
of competition regime if in the 1950s it had been subject to significant col-
lusive agreements covering at least 50% of total sales revenue and these
agreements were subsequently abandoned. It is defined as an industry
without a change in competition regime if less than 10% of the industry was
affected by the legislation. And it is defined as ambiguous (or the effect of the
1956 Act on the industry is defined as ambiguous) in all other cases.
My assessments below should be interpreted in relation to my classifica-
tion criteria. This should be kept in mind especially for cases where an in-
dustry is said to have experienced a ‘‘change in competition regime.’’ This
means that in the 1950s the industry in question had been subject to signifi-
cant collusive agreements covering at least 50% of total sales revenue and
these agreements were subsequently abandoned. In other words, the ex-
pression ‘‘change in competition regime’’ should not be taken to imply that
competition definitely intensified as a result of the 1956 Act in that particular
industry. Such information is available for several industries and will be
reported below; in many cases, however, information is simply not available
and the maintained assumption—which, as discussed in chapter 2, gets very
significant support from survey and case-study evidence—is that competi-
tion indeed intensified, sooner or later, in the large majority of the previously
collusive industries.

Stone and Slate Quarrying and Mining (MLH 102)

Stone mining and quarrying. About seventy-five regional agreements


in dry and coated stone for roadwork, building work, and so on were regis-
tered, then formally abandoned between 1959 and 1961. They all comprised
agreed prices, and some also provided for market sharing or the allocation
of work. There were also agreements between associations to respect each
Survey of Collusive Agreements in British Industries 337

other’s arrangements in their respective areas. These regional cartels proba-


bly covered more than half of total industry sales. At the national level, the
Roadstone Producers Advisory Committee gave recommendations for price
increases to cover changes in costs until 1961 (R1717). Swann et al. (1973)
provide details on this industry and also argue that competition gradually
intensified in the 1960s, although in some areas this had not yet occurred
by the late 1960s—partly as a result of information agreements. On the other
hand, an official investigation into the industry carried out in the 1970s
revealed a large number of regional or local restrictive agreements, many of
which had been in operation since the early 1970s, while others had been in
operation since the 1960s. These agreements were placed on the register (and
most were also referred to the Court), and were abandoned in the late 1970s.
All in all, the state of competition in the stone mining and quarrying industry
throughout the 1960s and the 1970s must be regarded as ambiguous.

Slate and slate products. A pricing agreement of the North Wales Slate
Quarries Association (R400) was abandoned in 1959. Wales accounted for
about 80% of the total UK production of slate and slate products in the 1950s.
Hence there was probably a change of competition regime in this industry.

Chalk, Clay, Sand, and Gravel Extraction (MLH 103)

Sand and gravel. About thirty regional agreements in sand and gravel
were registered, then formally abandoned between 1959 and 1961—although
some of them were replaced by information agreements. They all comprised
agreed prices, and some also provided for market sharing or the allocation
of work. There were also agreements between associations to observe each
other’s arrangements in their respective areas. The agreements probably
covered more than half of total industry sales; this is also the view expressed
in PEP. Three of the agreements (R595*, R943*, R1712*) were by Scottish
associations that had previously been the subject of an official inquiry—see
MRPC, Report on the Supply of Sand and Gravel in Central Scotland (London:
H.M.S.O., 1956). An official investigation into the sand and gravel industry
carried out in the mid-1970s revealed several dozen regional or local restric-
tive agreements which had been in operation since the early 1970s or even
earlier. These agreements were placed on the register (and most were re-
ferred to the Court), and were abandoned in the late 1970s. There is some
uncertainty about their coverage and effectiveness, however. In summary,
the state of competition in the industry during the 1960s and the 1970s
should probably be regarded as ambiguous.

Clay. Three national agreements covered all types of clay except fireclay.
The China Clay Association recommended price changes and required mem-
bers to report their individual prices and any intention to change them to
the association (R128). The China Stone Association fixed prices and quotas
(R190). And the Ball Clay Producers Federation’s agreement contained min-
338 Appendix A

imum list prices, although it did not regulate trade discounts (R377). All
three agreements were abandoned in 1959. In addition, various associations
of firebrick producers (listed under MLH 461 below) fixed the price of fire-
clay until 1959–1961. Overall, there was a change of competition regime in
the clay industry.

Other products. There is no evidence of any restrictive agreements.

Iron Ore and Other Metalliferous Mining and Quarrying


(MLH 109.1,2)

Iron ore. The case of iron ore is special in several ways. First, imports were
far greater than domestic production and they were centrally purchased.
Second, the Iron and Steel Board (the regulatory body for the steel industry)
had implemented a scheme that may have resulted in price discrimination
against users of domestic iron ore. Third, most of the domestic iron ore was
mined by steel firms, which were also the principal users. See D. Burn, The
Steel Industry 1939–1959 (Cambridge: Cambridge University Press, 1961) for
details. In any case, there is no evidence of any collusive agreements between
iron ore producers, and there was no change in competitive conditions be-
tween the mid-1950s and the mid-1960s. In 1967 the iron ore industry was
radically transformed as a result of the nationalization of steel.

Other metalliferous mining and quarrying. No evidence of any restrictive


agreements.

Salt and Other Nonmetalliferous Mining and Quarrying


(MLH 109.3,4)

Salt. The salt industry is mentioned as collusive in PEP, and it was also
listed in CD; however, no agreement was registered in the 1950s. An inves-
tigation of the industry by the MMC in the 1980s revealed an unregistered
pricing agreement, which had been in operation for several decades. See
MMC, White Salt (London: H.M.S.O., 1986). Thus the salt industry was col-
lusive throughout the time period examined in this book.

Other nonmetalliferous mining and quarrying. The British Fluorspar Pro-


ducers Association fixed prices for fluorspar until 1960 (R477).

Grain Milling (MLH 211)

Flour. Several agreements by various associations and groups of flour


millers were registered. Most of them were regional and provided for rec-
Survey of Collusive Agreements in British Industries 339

ommended prices (R175, R176*, R776*, R777, R778, R779, R780, R783*, R969,
R1247, R1599*, R1608, R1652*, R2276, R2429, R2443). There were also five
important national agreements. Three of them, by the National Association
of British and Irish Millers, covered most types of flour and associated
products, and comprised recommended prices and a scheme of aggregated
rebates for large buyers (R782*, R784*, R844). The fourth, by the Association
of Millers of Proprietary Brown Flour, required members to report their in-
dividual prices and any intention to change them to the association (R843).
The fifth, by the Millers Mutual Association, whose members included all the
important producers of flour, comprised a system of quotas (R2154). All the
national and regional agreements were abandoned in 1958–1959. Hence
there was a change of competition regime in the flour industry. An MMC
inquiry found competition in the industry to be effective in the 1970s—see
MMC, Flour and Bread (London: H.M.S.O., 1977).

Ready-to-eat breakfast cereals. No evidence of collusive agreements.

Other products. An agreement between members of the Lentil Millers


Association (R781), containing prices and quotas, and a pricing agreement
among producers of soy flour (R1495) were both canceled in 1959.

Bread and Flour Confectionery (MLH 212)

Bread. The price of bread was under government control until 1956. After
1956, it became subject to various agreements by national, regional, and local
associations. The three most important agreements were referred to the Re-
strictive Practices Court and were unsuccessfully contested by the parties
(see LR 1 RP 387, LR 1 RP 347). In particular, R962* was a pricing agreement
by the Federation of Wholesale and Multiple Bakers, whose members were
‘‘plant bakers’’ in England and Wales. R252*, R981*, and R980* were price
agreements by the Wholesale and Retail Bakers of Scotland Association, the
Scottish Association of Master Bakers, and a Joint Costing Committee of the
two associations, respectively. All these agreements were abandoned in
1959–1960. Regional agreements in the industry (R985, R1048, R1311, R2462,
R2469, R2470, R2471, R2478, R2480, R2580, R2599, R2601) were also termi-
nated at that time, as were agreements by associations of ‘‘master bakers’’
(R1542, R1566, R2275) and numerous local collusive arrangements.
Swann et al. (1973) describe the gradual emergence of competition in the
bread industry, despite parallel pricing and some information agreements
about discounts—for instance, R3061*, which ended in 1965. Further infor-
mation is contained in MMC, Flour and Bread (London: H.M.S.O., 1977). The
MMC discovered a large number of agreements between leading firms that
had been in operation since the late 1960s or early 1970s. Most of them
involved the exchange of information about discounts to particular buyers
or about intended increases in discounts. Some arrangements were national,
340 Appendix A

but most were local. All of them were placed on the register (and some were
referred to the Court) and were abandoned in the mid-1970s. See the report
of the Director General of Fair Trading for 1977 for details. However, it is
very doubtful that these arrangements had any significant effect on compe-
tition. The view of the MMC was that they were an attempt, largely unsuc-
cessful, to resist pressure from retailers for progressively larger discounts.
In fact, discounts kept rising between 1965 and 1975, which suggests that
competition was largely effective. Hence the bread industry experienced
a change of competition regime as a result of the 1956 Act.

Flour confectionery. This industry was not subject to any national agree-
ments. Some of the regional or local pricing arrangements among bakers
covered flour confectionery as well, but most did not. These agreements
probably affected more than 10% but less than 50% of total industry sales.

Biscuits (MLH 213)

An agreement by the National Association of Biscuit Manufacturers provided


for prices, conditions of sale, and restrictions on sales promotion (R1519); it
was abandoned in 1959. Chocolate biscuits were also covered by agreements
between chocolate confectionery manufacturers; they ended in the early 1960s
(see MLH 217 below). The biscuit industry was also listed in CD, which
implies that it was subject to collective discrimination arrangements before
1956. In summary, there was a change of competition regime in this industry.
Cereal filling, a secondary product of MLH 213, was also subject to a pricing
agreement, which was operated by the Rusk Manufacturers Association and
was abandoned in 1960 (R610*).

Bacon Curing, Meat and Fish Products (MLH 214)

Bacon. The British Bacon Agents Association recommended prices for Brit-
ish bacon (R2750). Its members were agents appointed by curers or curers
who acted as their own agents. The agreement ended in 1961 and was re-
placed by an information agreement, which lasted until the early 1970s.
However, about a third of the members withdrew from the association in
1962–1963. A pricing agreement by the Association of Scottish Bacon Curers
and the Scottish Provision Trade Association was abandoned in 1960
(R1185*). Finally, the register also contains a series of annual contracts for the
purchase of pigs between farmers’ representatives and the British Bacon
Curers Federation or smaller associations of curers (R2301, R2305, R2286).
These contracts, which specified prices and terms of purchase, had been a
feature of the industry ever since the end of government control of the dis-
tribution of fatstock, and continued until the early 1970s. It is unlikely that
they had any significant effect on competition between producers of bacon.
In conclusion, there was a change of competition regime in this industry.
Survey of Collusive Agreements in British Industries 341

Sausages and cooked meats. An agreement by the Scottish Association of


Cooked Meat Manufacturers, which ended in 1960, contained recommended
price changes for sausages and other meat products (R99). However, Scottish
producers accounted for less than 10% of the total UK production of sau-
sages and cooked meats, so the industry was largely competitive; this is
consistent with the evidence reported in PEP.

Other meat and fish products. An agreement by the Association of British


Salted Fish Curers and Exporters, abandoned in 1959, related only to maxi-
mum prices for the purchase of imported wet salted fish (R241). An agree-
ment between members of the Salmon Merchants Association of Great
Britain, who were curers of raw salmon, was also about buying prices and
was abandoned in 1959 (R473).

Milk and Milk Products (MLH 215)

Heat-treated milk. A number of regional and local agreements among


processors/distributors of milk were registered, then abandoned between
1959 and 1965. Most of these were market-sharing agreements, but some also
contained pricing restrictions. Several were referred to the Court (R2213*,
R1222*, R2988*, R2119*, R2198*, R2199*, R2574*). The fraction of the total UK
production of heat-treated milk covered by these agreements is difficult to
estimate; it may have been less than 50%. The industry was also included in
the BT lists of allegedly collusive industries, but no precise information was
given on coverage. Adding to the uncertainty regarding the evolution of
competition in this industry is the fact that maximum prices and profit mar-
gins were regulated by the Milk Marketing Board (the regulatory body for
the milk industry) at least until the late 1960s. It is therefore clear that the
effect of the 1956 Act on this industry should be seen as ambiguous.

Butter. The register contains two national pricing agreements: R814 by the
English Butter Conference and R1047 by the Butter Makers and Packers As-
sociation. Both ended in 1959–1960. Hence there was a change of competi-
tion regime in the butter industry.

Cheese. The cheese industry comprises two different product categories:


processed cheese, which accounted for less than 50% of total industry sales
throughout the period, and nonprocessed cheese. There is no evidence of any
collusive agreements in processed cheese. In contrast, several types of non-
processed cheese were subject to collusive pricing. In particular, the Cheshire
Cheese Makers Association (R396), the Lancashire Cheese Association (R813),
the Company of Scottish Cheesemakers (R285), the Wensleydale Cheese Joint
Conference (R580), and the Cheddar and Caerphilly Cheese Makers Asso-
ciation (R424, R2558) recommended or fixed prices of cheese. All these
agreements ended in 1959–1960. R396 was replaced for some time by an in-
342 Appendix A

formation agreement. The fraction of the cheese industry affected by the 1956
Act was probably around 40%.

Cream. The National Association of Creamery Proprietors and Wholesale


Dairymen recommended minimum prices for fresh cream (R812). The Clotted
Cream Makers Association recommended price increases for clotted cream
to cover changes in costs (R395). Finally, the Association of Tinned Cream
Manufacturers recommended minimum prices of tinned sterilized cream
(R582). All these arrangements were canceled in 1959, leading to a change of
competition regime in the cream industry.

Condensed milk. The Association of British and Dominion Condensed Milk


Manufacturers recommended minimum prices for all classes of condensed
milk until 1960 (R585). The abolition of this agreement led to a change of
competition regime in the industry.

Milk powder. The Association of British Manufacturers of Milk Powder


recommended minimum prices for milk powder until 1960 (R584, R421).
Thus the industry experienced a change in competition regime.

Ice cream. No evidence of any restrictive agreements.

Other milk products. An agreement by the Imitation Cream Division of


the Bakery Allied Traders Association (R2567) contained recommendations
about price changes; it was abandoned in 1960.

Sugar (MLH 216)

Prices, market shares, and profit margins in the sugar industry were under
government regulation throughout the period examined in this book. Details
can be found in Hart et al. (1973) and in MMC, S & W Berisford Limited and
British Sugar Corporation Limited: A Report on the Proposed Merger (London:
H.M.S.O., 1981). All the registered agreements relating to sugar were among
distributors, except one, which was a pricing scheme implemented by the
government and the three main manufacturers in 1974 (R3833). It replaced
previous schemes in the industry and was discontinued in 1976.

Cocoa, Chocolate, and Sugar Confectionery (MLH 217)

Cocoa and chocolate. The register contains a series of agreements between


five (later four) of the largest producers in the industry, covering cocoa,
drinking chocolate, chocolate confectionery, and chocolate biscuits (R287*,
R2639*, R288). For the first three of these product categories, the restrictions
Survey of Collusive Agreements in British Industries 343

included maximum discounts to distributors, the requirement that each firm


enforces its (individually set) retail prices, the prenotification of any changes
in individual prices or discounts, and numerous provisions regarding sales
promotion. For chocolate biscuits, there were only restrictions on sales pro-
motion. All the arrangements ended between 1960 and 1962.
Another agreement, by the Wholesale Confectioners Alliance of Great
Britain and Northern Ireland (R169*), was referred to the Court in 1961 and
was unsuccessfully contested by the parties (LR 2 RP 135, LR 2 RP 231). This
was an association of wholesalers of chocolate and sugar confectionery. The
agreement essentially involved an agreed split between wholesalers and
retailers of the total difference between producer prices and consumer prices
(which were fixed and maintained by the individual manufacturers). The
effect of this on competition between manufacturers is somewhat uncertain,
but at least it implies that resale price maintenance and the restrictions on
trade discounts were effective throughout the industry. Furthermore, the
existence of an effective price notification agreement is confirmed by PEP,
and the industry was also listed in CD, which confirms the existence of col-
lective discrimination arrangements before 1956. In conclusion, the industry
probably experienced a change of competition regime as a result of the 1956
Act.

Sugar confectionery. The agreement of the Wholesale Confectioners


Alliance of Great Britain and Northern Ireland (R169) related to sugar
confectionery as well. In this case, however, the arrangements between man-
ufacturers provided only for restrictions on sales promotion and the mainte-
nance of individual resale prices (see R287, R2639, R288, and the Lloyds’
report). The industry was also listed in CD. The state of competition in this
industry in the 1950s should probably be regarded as ambiguous.

Fruit and Vegetable Products (MLH 218)

There is no evidence of any significant restrictive agreements in this industry.

Animal and Poultry Foods (MLH 219)

Animal feeding stuffs, other than dog and cat foods. No important agree-
ments were registered in the 1950s. (An agreement by the Association of Fish
Meal Manufacturers, R1635, related only to delivery charges; besides, fish
meal is a minor product of the industry). However, the industry was listed
in BT as allegedly collusive. In addition, an official investigation carried
out in the 1970s revealed that a collusive agreement between leading pro-
ducers had been in operation since 1970 (R4556*, R4673*). This was aban-
doned in 1978. All in all, the effect of the 1956 Act on this industry was
ambiguous.
344 Appendix A

Dog and cat foods. The National Association of Dog Biscuit Manufacturers
fixed the price of biscuit powder or dust, biscuit meals, and dog meals (R96).
The agreement dealt only with residuals and low-priced dog and poultry
food, not with the manufacturers’ proprietary brands. Since the whole of the
product category ‘‘dog biscuits’’ (including proprietary brands) accounted
for only about 10–20% of total sales of dog and cat foods during most of the
relevant period, the industry can be seen as essentially competitive.

Vegetable and Animal Oils and Fats (MLH 221)

A pricing arrangement by the Scottish Association of Edible Fat Melters


(R1186) ended in 1959. PEP mentions the Raw Fat Melters Association as al-
legedly collusive. Also, both BT and PEP mention an arrangement whereby
the members of the Federation of Bone Users and Allied Trades regulated the
prices and allocated the supplies of fat and bones for the production of tal-
low, grease, and other products (in this case the information in PEP comes
from a firm in the industry). The combined share of the products covered by
these associations was less than 10% of total industry sales revenue. Thus
both MLH 221 as a whole and most of its principal products experienced no
change in competition regime as a result of the 1956 Act.

Margarine (MLH 229.1)

There is no evidence of any restrictive agreements in this industry apart from


an allegation about collective discrimination arrangements in margarine
reported in BT.

Starch and Miscellaneous Foods (MLH 229.2)

Starch and glucose. The British Dextrine Manufacturers Association rec-


ommended minimum prices for all dextrins until 1959 (R242). The affected
products accounted for about 20% of sales revenue in this product group.
Moreover, liquid glucose is mentioned in PEP and BT as allegedly collusive.

Coffee, and coffee and chicory extracts and essences. No evidence of any
restrictive agreements.

Self-raising flour. The Self-Raising Flour Association recommended prices


for self-rising flour until 1959 (R185). There was therefore a change of com-
petition regime in this industry.

Other products. No evidence of any restrictive agreements.


Survey of Collusive Agreements in British Industries 345

Brewing and Malting (MLH 231)

Beer. Several brewers took part in regional price-fixing agreements by


wholesale bottlers, which were generally abandoned in 1958–1959. The reg-
ister also contains some regional agreements by groups of brewers. These
regulated the relations between brewers and public houses, and sometimes
also provided for the prenotification of individual price changes, but did not
impose restrictions on individual prices (R45, R1457, R1602, R2239, R2317).
The effect of these arrangements on competition between beer producers
is difficult to assess. Moreover, the fraction of the UK beer market covered
by these agreements, which remained in force until the mid-1960s, was
probably between 20% and 50%.

Malt. An annual agreement between various associations of farmers,


maltsters, and brewers regarding the price of barley destined for malting
was not renewed after 1959 (R1589, R1849, R2382). An agreement by the
Maltsters Association of Great Britain, relating to conditions of sale, was
partly canceled in 1967 (R199). These are unlikely to have had any significant
effect on competition between malt producers, so the industry can be seen as
competitive throughout the period.

Soft Drinks (MLH 232)

The register contains many regional price-fixing agreements by various as-


sociations of manufacturers/bottlers (R127, R267, R270, R589, R594, R738,
R794, R982, R1162, R1199, R1229, R1250, R1300, R1419, R1709, R2284). All
ended between 1958 and 1960. The fraction of the total UK soft drinks market
affected by these regional cartels is difficult to estimate, because the agree-
ments did not cover national proprietary brands. Also, according to PEP, the
agreements were not always effective. As far as national brands are con-
cerned, an arrangement between two leading firms, Schweppes and Lyons,
which may have amounted to market sharing, remained in force until the
1970s, because of various technical reasons relating to the form of the agree-
ment (R3679, R3680). See the Registrar’s report for the period 1969–1972 and
the annual report of the Director General of Fair Trading for 1974 for details.
Overall, the effect of the 1956 Act on the soft drinks industry should be
seen as ambiguous, although the effect on competition between the large
advertising-intensive firms was probably not significant.

Spirit Distilling and Compounding (MLH 239.1)

Spirit distilling. An agreement between members of the Pot Still Malt


Distillers Association, relating to prices of malt whiskey sold to blenders,
was abandoned in 1956 and was never registered. The information comes
both from PEP and from the October 1956 issue of the periodical Cartel.
346 Appendix A

According to PEP, there were no similar arrangements among grain dis-


tillers, but intentions to change individual prices were notified. In conclu-
sion, the industry probably experienced a change in competition regime
after 1956.

Whisky. Domestic and export sales of whiskey were officially rationed until
1954. In fact, rationing in the home market continued until the late 1950s (see
Hart et al. 1973). Moreover, the firms agreed on maximum prices for whiskey,
according to CD, PEP, and BT, although no such agreement appears in the
register. The effect of rationing and of maximum prices on competition is
difficult to determine in general, so the state of competition in this industry
in the 1950s should probably be seen as ambiguous.

Other products. No evidence of any significant restrictive agreements


affecting manufacturers. ‘‘Wines and spirits’’ are listed in CD.

British Wines, Cider, and Perry (MLH 239.2)

There is no evidence of significant restrictive agreements affecting manu-


facturers. ‘‘Wines and spirits’’ are listed in CD.

Tobacco (MLH 240)

The principal sources of information on the tobacco industry are PEP (which,
in this case, reports evidence given by a tobacco firm), the Lloyds’ report,
and especially MC, Report on the Supply of Cigarettes and Tobacco and of Ciga-
rette and Tobacco Machinery (London: H.M.S.O., 1961). According to these
sources, the most important factor restricting competition in the industry
during the early 1950s was the government-controlled allocation of raw
materials among producers. This ended in 1954, and was followed by a large
expansion in the market share of the two leading firms at the expense of
smaller firms. An agreement by the Northern Tobacco Manufacturers Asso-
ciation to set minimum prices of cheaper types of tobacco sold in Scotland
and northern England also ended at that time; in any case, the largest firm in
the industry had never been a party to that arrangement. After 1954, there
were no horizontal agreements on prices, although the industry practiced
collective resale price maintenance through the Tobacco Trade Association
until 1956 (see also CD and BT). There were also occasional discussions
about distributors’ margins, but no explicit agreement. The register does not
contain any important restrictive agreements among tobacco manufacturers.
This evidence suggests (1) that the state of competition in the industry was
ambiguous until the mid-1950s because of government control, and (2) that it
would probably be safe to classify the industry as one without a change in
competition regime between the late 1950s and the 1970s.
Survey of Collusive Agreements in British Industries 347

Coke and Manufactured Fuel (MLH 261)

Coke. The register contains two national and several regional agreements
on coke. The price of blast-furnace coke was regulated by an agreement
between the British Coking Industry Association, the British Iron and Steel
Federation, and the Council of Iron Producers (R649*). The British Coking
Industry Association also recommended prices of hard coke and had mar-
keting schemes for various regions (R650*). Both agreements ended in 1964–
1965. Further information on the hard coke industry is provided in PEP,
in which it is pointed out that the retail prices of hard coke were regulated
by the government, while the producer prices fixed by the British Coking
Industry Association were implemented as part of a government-approved
scheme. What is important for our present purposes is that during 1954–
1963, there was no change in the state of competition in the coke industry
(hard coke and blast-furnace coke). After the mid-1960s, conditions changed
dramatically as a result of the nationalization of the steel industry, since the
steel firms produced about half of the total output of coke. The register also
contains a series of price agreements for gas coke, which were abandoned in
the early 1960s. However, gas coke is not part of MLH 261.

Manufactured fuel. No evidence of restrictive agreements.

Other products. Several agreements in the register related to schemes


whereby a number of producers of crude tar entered into similar contractual
arrangements with a single refiner, specifying the prices to be charged,
amounts to be sold, and so on. Nearly all of these ended between 1958 and
1962.

Mineral Oil Refining (MLH 262)

Several agreements in the register related to the common marketing of vari-


ous mineral oil refining products by two (initially three) of the largest firms
in the industry, Shell and BP (R1139, R3717, R1150, R1350, R1708). This
continued until 1976. Further information on the industry is contained in
PEP, BT, and MC, Petrol (London: H.M.S.O., 1965). Several principal prod-
ucts of MLH 262, such as petrol (gasoline), fuel oil, and diesel oil, were listed
as allegedly subject to collusive pricing and other restrictions in BT and/or
PEP, and the industry was also listed in CD. The Monopolies Commission took
a more favorable view on the industry, arguing that there was no evidence of
price-fixing in petrol after 1953, despite the existence of parallel pricing. (Before
1953, the price of petrol was under government control.) Instead, there was
intense competition for outlets, partly through price incentives to retailers.
Moreover, the MC did not consider the common marketing arrangements
between Shell and BP to be significant restrictions on competition. Finally,
348 Appendix A

the MC noted that price competition intensified after 1960, but it attributed
this to the entry of new firms. All in all, the effect of the 1956 Act on com-
petition in the mineral oil refining industry was ambiguous.

Lubricating Oils and Greases (MLH 263)

The register contains agreements about the common marketing of lubricating


oils by two (initially three) of the largest firms in the industry, the same as in
MLH 262 (R1133, R3718). This continued until 1976. A price agreement by
the White Oils Association, covering certain types of lubricating oils, ended
in 1960 (R2161). Very few of the members of this association were refiners;
most were distributors who cleaned, mixed, and packed the oils purchased
from the large refiners. Thus the agreement was of limited relevance for
competition among manufacturers. Also, the products covered accounted for
a small fraction of total industry sales. Hence the 1956 Act probably had no
significant effect on this industry.

General Chemicals, Inorganic (MLH 271.1)

An arrangement for the common buying of sulfur by members of the Na-


tional Sulphuric Acid Association was referred to the Court, contested by the
parties, and allowed to continue (R123*—see also LR 4 RP 169, LR 6 RP 210).
It remained in force until at least the late 1970s, but its effect on competition
among producers of sulfuric acid is not clear—partly because about half of
the total output of sulfuric acid was produced from other raw materials and
partly because the agreement was a response to the existence of an export
cartel among foreign suppliers of sulfur. On the other hand, according to a
case study of the chemical industry, written by W. B. Reddaway and pub-
lished in Burn (1958), some regional sections of the association had agree-
ments on selling prices before 1956. None of these was registered.
There are four other agreements under MLH 271.1 in the register. The
producers of copper sulfate had a common selling organization, the British
Sulphate of Copper Association (R134). Their agreement, which prescribed
that the parties should sell all their production through the company, ended
in 1960. The Potassium Carbonate Association regulated the purchase and
distribution in the UK of potassium carbonate until 1959 (R785). The Scottish
Soda Crystal Manufacturers Association operated as a common selling
agency for the soda crystal produced by its members (R1331). Until 1959, it
fixed prices and required that members sell all their production through the
association. On the other hand, an agreement by the British Acetylene Asso-
ciation related only to conditions of sale and was allowed to continue with-
out a Court hearing, since it was not regarded as restrictive (R1959). Finally,
two other products of MLH 271.1 are listed in BT as potentially collusive:
sodium metasilicate, and salt and sodium sulfide. The fraction of total in-
Survey of Collusive Agreements in British Industries 349

dustry sales in MLH 271.1 accounted for by collusive or allegedly collusive


products was probably around 25–35%.

General Chemicals, Organic (MLH 271.2)

Products of coal tar distillation. Almost all the products in this industry
were subject to collusion. The Association of Tar Distillers recommended
prices for road tar and bitumen tar mixtures, coal tar fuels, and certain cre-
osote oils (R1033). The agreement was canceled in 1960 and replaced by an
information agreement, which ended in 1965. Refined tar was also subject to
regional agreements until 1960 (R1417, R1458). The Anthracene Producers
Committee of the Association of Tar Distillers fixed prices of anthracene until
1965 (R3126, R3127). Prices of creosote oil for hydrogenation were regulated
until 1958 (R3153). Crude tar acids were subject to various pricing agree-
ments between tar distillers and refiners until 1961–1962 (R222, R753, R177,
R178, R179). Finally, sales of briquetting pitch were regulated by a common
selling organization, the Pitch Pool of the Association of Tar Distillers, until
1974 (R3154). All in all, there was a change of competition regime in the long
run in this part of MLH 271.2.

Other organic chemicals. The Phenol Producers Association fixed the


prices of phenol, orthocresol, and phenol-cresol mixtures (R219*, R220).
R219* was contested by the parties in the Court and was struck down (LR 2
RP 1). Both agreements were abandoned in 1960, although R219* may have
been replaced for some time by an information agreement. The Cresylic Acid
Refiners Committee recommended minimum prices for cresylic acid and var-
ious mixtures until 1960 (R221). Crude and refined benzole (benzene) were
sold through a common selling organization, Benzole Producers (R1025,
R2081, R3090). The agreement, which prescribed that the parties should sell
all their production through the company, continued until the 1970s. Finally,
the Naphthalene Producers Committee of the Association of Tar Distillers
recommended prices of naphthalene for standard buyers until 1961 and fixed
prices for large buyers until 1966 (R2826, R2678, R2679). All these products
accounted for a small fraction of total sales in this part of MLH 271.2, how-
ever, so the effect of the 1956 Act on competition was not significant.

General Chemicals, Other Than Inorganic and Organic (MLH 271.3)

A pricing agreement between producers of case-hardening compounds


ended in 1959 (R62). An agreement between two sellers and five buyers of
petroleum-cracking catalysts, providing for cooperation in production, joint
sale, and exclusive supply arrangements, was canceled in 1962 (R1502,
R1521). These are very minor products of MLH 271.3, which can therefore be
regarded as a competitive industry.
350 Appendix A

Pharmaceutical Chemicals and Preparations (MLH 272)

Pharmaceutical chemicals. A restrictive agreement between two producers


of insulin (R950), providing for cooperation in production and for the joint
sale of insulin, was canceled in 1961. See also MRPC, Report on the Supply of
Insulin (London: H.M.S.O., 1952). Insulin is a very small part of the industry,
which was therefore essentially competitive.

Pharmaceutical preparations. The agreement of the Chemists Federation


(R1027*) was the first to be referred to the Restrictive Practices Court and to
be contested by the parties in late 1958. The membership of the federation
consisted of manufacturers of proprietary medicines, wholesale chemists,
and retail chemists. The agreement, which imposed restrictions designed to
prevent the sale to the public by anyone except retail chemists (whether or
not members) of proprietary medicines manufactured by members, was
condemned by the Court (LR 1 RP 75). However, it is not clear why this
agreement should have had any significant effect on competition between
manufacturers. The same remark can be made about the activities of the
Proprietary Articles Trade Association, whose membership consisted of
manufacturers and distributors of medicines and toiletries: the association
collectively enforced resale prices and set minimum distributors’ margins
until 1956, but it did not regulate the manufacturers’ individual prices. See
the Lloyds’ report for details (the industry was also listed in CD). An agree-
ment of the Proprietary Association of Great Britain, which remained in force
throughout the period examined in this book, related only to a code of con-
duct (R3259). On the other hand, a pricing agreement between three UK and
several foreign manufacturers of quinine and quinidine (R3111, R3112),
which ended in 1963, covered a very specific product and was of little im-
portance for competition in the industry as a whole.
In contrast, an agreement by the Association of the British Pharmaceutical
Industry, which ended in 1960, contained recommended prices for non-
proprietary medicines for the home retail and hospital trade (R1134). The
agreement did not affect (1) branded proprietary drugs and (2) export sales.
According to a case study of the industry, written by C. J. Thomas and pub-
lished in Burn (1958), roughly a third of total sales of drugs in the mid-1950s
were export sales; another third were home sales of branded and unbranded
drugs through the National Health Service (either to hospitals or supplied on
prescription), and about 70% of those were sales of branded drugs; another
third were publicly advertised proprietary medicines, drugs sold without
prescription, and veterinary and horticultural medicines. These figures sug-
gest that sales of unbranded drugs in the home market, either through the
NHS or without prescription, did not account for more than 10–15% of total
industry sales revenue in the mid-1950s, and they probably accounted for
less than 10% in the 1960s. Since the only products affected by the agreement
of the Association of the British Pharmaceutical Industry were unbranded
Survey of Collusive Agreements in British Industries 351

drugs sold in the home market, the industry should probably be seen as
largely competitive.

Toilet Preparations (MLH 273)

A registered agreement by the Hairdressing Manufacturers and Wholesalers


Association (R482) specified conditions of sale and minimum trade discounts
for hairdressers’ articles. Most of the restrictions were abandoned in 1959.
According to the Lloyds’ report, this association was also responsible for the
collective enforcement of manufacturers’ individual resale prices (until
1956). Similarly, the Proprietary Articles Trade Association enforced resale
prices and set minimum distributors’ margins for other toilet preparations
until 1956. The industry was also listed in CD. It is not clear why these
arrangements would have had a significant effect on competition between
manufacturers. (Note that the membership of both associations consisted of
manufacturers and distributors.) Thus both the industry as a whole and all
its principal products should be seen as competitive.

Paint (MLH 274)

Paint. Each of the two associations in the paint industry, the Society of
British Paint Manufacturers and the National Federation of Associated Paint,
Colour and Varnish Manufacturers of the U.K., registered a restrictive agree-
ment (R1920, R1896). A third agreement was registered by the British Paint
Advisory Council, which had been set up by the two associations (R385). All
three agreements provided for the maintenance of the manufacturers’ (indi-
vidually set) resale prices, and specified trade discounts for various classes of
buyers and various types of products, quantity discounts, and terms and
conditions of sale. The industry was also listed in CD. The agreements were
canceled in 1959. Although the evidence from PEP suggests that these
agreements may have not been fully effective, the paint industry should be
regarded as one with a change of competition regime.

Other products. A price-fixing agreement by the National Association of


Putty Manufacturers (R643), relating to putty and building mastics, was
canceled in 1959. It was replaced by an information agreement, which was
abandoned in the early 1970s.

Soap and Detergents (MLH 275)

Soap. No evidence of restrictive agreements.

Detergents. No evidence of restrictive agreements in the 1950s. An agree-


ment by the two leading firms, Unilever and Procter and Gamble, was in
352 Appendix A

operation between 1960 and 1968 and was essentially an attempt to resist
an ongoing trend toward higher promotional expenditure in the industry
(R2983). According to MC, Household Detergents (London: H.M.S.O., 1967),
promotional expenditure was still excessive in the mid-1960s, although it
may have declined subsequently. In any case, there is no evidence of any
significant change in the state of competition in the industry between the
1950s and the 1970s.

Other products. An agreement by the U.K. Glycerine Producers Associa-


tion comprised prices for crude and refined glycerine (R640*, R641). It was
abandoned in 1960.

Synthetic Resins and Plastics Materials, and Synthetic Rubber


(MLH 276)

Synthetic resins and plastics materials. The register contains evidence of


informal discussions about prices by manufacturers of polyester resin in the
1970s (R5043). Moreover, polystyrene molding powder was listed in BT as
allegedly collusive in the 1950s. The share of these products in total industry
sales revenue was probably less than 10%.

Synthetic rubber. No evidence of restrictive agreements.

Cellulose film. No registered agreements, although this product was listed


in BT as allegedly collusive.

Dyestuffs and Pigments (MLH 277)

Dyestuffs. No evidence of any major restrictive agreements. The British


Tanning Extract Manufacturers Association set the prices of tanning extracts
sold to the leather industry until 1959 (R755). This is a very minor product,
however, so the dyestuff industry as a whole was not significantly affected
by the 1956 Act.

Pigments. The register contains several agreements covering specific prod-


ucts. The U.K. White Lead Convention fixed the price of white lead (R786).
The prices of zinc oxide (R2045*), aluminum red oxide (R2479), ‘‘helio fast red’’
pigment (R555), ‘‘persian gulf red’’ pigment (R592), and various chromium
pigments (R756) were also fixed. All these agreements were abandoned
in 1959–1960. An international market-sharing agreement in lithophone,
involving two UK firms, ended in 1964 (R2989). The combined share of all
these products in total industry sales revenue was about 25% in the 1960s.
Survey of Collusive Agreements in British Industries 353

Finally, according to CD, lead oxide was subject to certain forms of collective
discrimination, such as aggregated rebates, before 1956.

Fertilizers (MLH 278)

The register contains three agreements. The British Sulphate of Ammonia


Federation fixed the price of sulfate of ammonia, in agreement with one large
producer outside the federation, until 1963 (R415*, R416*). A selling organi-
zation for basic slag, British Basic Slag, had a series of bilateral agreements
with its parent firms, requiring each of these firms to sell the whole of its
production of basic slag through the joint company (R1024*, R2865*). The
Court recognized that this was effectively an arrangement between the slag
producers, and was therefore registrable under the 1956 Act (LR 3 RP 178,
LR 4 RP 116); it was abandoned in 1964. Finally, the Phosphate Rock Agency,
a common buying organization for phosphate rock, registered and later
abandoned an agreement specifying that members should purchase phos-
phate rock only through the agency (R132). The products affected by these
agreements accounted for about 10% of total sales of fertilizers.
Additional information on the industry is contained in MC, Report on the
Supply of Chemical Fertilisers (London: H.M.S.O., 1959). The price of fertilizers
was under government control until 1953. At that time, firms in the industry
were charging common prices and the two principal associations, the
Superphosphate Manufacturers Association and the Fertiliser Manufacturers
Association, recommended trade discounts and conditions of sale. Accord-
ing to CD, the industry was also subject to collective discrimination arrange-
ments. There was no agreement on common prices after 1953, however, and
all recommendations on discounts and conditions of sale ceased in 1956. In
conclusion, the state of competition in the fertilizer industry was ambiguous
in the early 1950s because of government regulation; in the late 1950s, only a
fraction of the industry was subject to collusive agreements.

Polishes (MLH 279.1)

There is no evidence of restrictive agreements in this industry.

Formulated Adhesives, Gelatine, Glue, Size, etc. (MLH 279.2)

In a series of agreements between its members and with associations in the


leather industry, the Hide, Gelatine, and Glue Section of the Federation of
Gelatine and Glue Manufacturers regulated the buying price of gluestock
and probably also allocated supplies among its members (R538, R1181,
R1182). The buying price of shredded rabbit pelt was also regulated (R1183).
The details of the allocation scheme for gluestock are not clear. On the other
hand, both BT and PEP mention an apparently different agreement where-
by the members of the Federation of Bone Users and Allied Trades regulated
354 Appendix A

the prices and allocated the supplies of fat and bones for the production of
glue and other products. The products affected by all these arrangements
accounted for about 15% of MLH 279.2 in 1963, but for less than 10% in 1968
and in the 1970s. Because of a substantial change in the definition of the in-
dustry between the 1958 S.I.C. and the 1968 S.I.C., the pre-1963 figures are
not comparable with the post-1963 figures. Hence it is sufficient, for our
present purposes, that the industry can be seen as one without a change in
competition regime after 1963.

Explosives, Fireworks, and Matches (MLH 279.3)

Explosives. No evidence of restrictive agreements.

Fireworks. An agreement of the British Pyrotechnists Association required


members to enforce their (individually set) resale prices and specified trade
discounts and conditions of sale (R561). The agreement was canceled in 1959,
and the association was dissolved in 1962. An agreement by its successor, the
British Firework Manufacturers Safety Association, which lasted until 1970,
related only to conditions of sale and safety standards (R3214, R3591). Hence
there was a change of competition regime in the fireworks industry.

Matches. There is no evidence of restrictive agreements between UK firms,


although there was an arrangement, involving prices and quotas, between
the dominant UK firm and a foreign competitor. See MRPC, Report on the
Supply and Export of Matches and the Supply of Match-making Machinery
(London: H.M.S.O., 1953).

Formulated Pesticides and Disinfectants (MLH 279.4)

There is no evidence of significant restrictive agreements in this industry.

Printing Ink (MLH 279.5)

The register contains an agreement by the Society of British Ink Manu-


facturers, relating to settlement discounts for sales of printing inks other than
news inks (R1376). The agreement also made vague references to attempts by
the association to raise the level of net prices by reducing discounts, and the
industry was also mentioned as potentially collusive in PEP. It is perhaps
best to consider the ink industry as ambiguous.

Surgical Bandages, etc., and Sanitary Towels (MLH 279.6)

Surgical and medical dressings. The Surgical Dressings Manufacturers


Association fixed prices for various types of dressings (R149). This formally
Survey of Collusive Agreements in British Industries 355

ended in 1959, but the member firms established an information agreement


and subsequently went through a prolonged period during which price wars
alternated with renewed, and partly successful, attempts to collude (R3213,
R3608). These attempts seem to have ceased after 1969. Swann et al. (1973)
provide full details. A second association in the industry, the Medical and
Surgical Plaster Makers Conference, fixed prices for adhesive bandages and
dressings until 1959 (R1078). According to CD, the surgical dressing industry
was also subject to collective discrimination arrangements before 1956. In the
long run, then, the industry experienced a change in competition regime.

Sanitary towels. No evidence of restrictive agreements.

Photographic Chemical Materials (MLH 279.7)

A price agreement between members of the Federation of Engineers’ Sensi-


tised Material Manufacturers ended in 1959 (R554). It was replaced by an
information agreement, which lasted until 1966. The products covered in the
arrangement, sensitized paper and cloth, accounted for about 30% of total
industry sales. However, an official investigation into the industry carried
out in the 1970s revealed several collusive agreements on reprographic
chemicals, diazo copying materials, and drafting film, which had been in
operation since the late 1960s or the early 1970s. See MMC, Diazo Copying
Materials (London: H.M.S.O., 1977) for details. These agreements, which
again covered only a fraction of MLH 279.7, were placed on the register (and
were also referred to the Court), and were abandoned in the late 1970s. In
summary, the state of competition in the industry throughout the period
examined in this book can only be classified as ambiguous.

Iron and Steel, General (MLH 311)

Wrought iron. The British Wrought Iron Association recommended prices


for wrought iron until 1965 (R380*). This is a very minor product of MLH
311.

Ferromanganese. The register contains a number of pricing agreements


involving producers of ferromanganese (R1232, R2049*, R2773*). R1232
ended in 1958, and the other two were abandoned in 1963. This is a secondary
product of MLH 311.

Steel products. The register contains more than seventy price-fixing agree-
ments by various associations and groups of steel producers, covering virtu-
ally every product of the steel industry. Some of these products also appear
in CD. Rather than listing the individual agreements, which would be very
356 Appendix A

tedious, I will provide here a summary for the industry as a whole and for its
principal products.3
One of the principal functions of the Iron and Steel Board (the regulatory
body for the steel industry) was to set maximum prices for a wide range of
steel products. For products covered by these controls, the agreements of the
steel producers essentially specified or recommended that the maximum
prices set by the Board be treated as fixed prices. For products free of gov-
ernment controls, the prices were set by the relevant associations. The
agreements typically specified trade discounts and conditions of sale as well.
A few also provided for the reporting of inquiries received from customers,
which could sometimes lead to the sharing of orders.
Most agreements in the steel industry were referred to the Court, and two
were contested by the parties but struck down by the Court. The first
(R1092*) was a pricing agreement by three associations representing manu-
facturers of heavy steel plates, sections, and joists—about 25% of all steel
products (LR 5 RP 33). The second (R1091, R2445*) was between the National
Federation of Scrap Iron, Steel and Metal Merchants and the British Iron and
Steel Federation, and imposed pricing and other restrictions on the market for
steel scrap (LR 4 RP 299). While the vast majority of the registered agreements
were between British firms, the register also contains evidence that the British
Iron and Steel Federation participated in an international restrictive agree-
ment on steel products, at least during 1962–1965 (R2835).
Most agreements were terminated in 1964–1965, but some were abandoned
before 1963. The most important products covered by agreements abandoned
before 1963 were steel castings, steel forgings, and tinplate. All the other
products were covered by agreements canceled after 1963. This distinction is
important in the present context because the steel industry was nationalized
in 1967; hence the only relevant period for our present purposes is the period
1954–1963. Whenever an agreement ended after 1963, the product in ques-
tion can be classified as one that experienced no change in competitive con-
ditions over the relevant period. Whenever an agreement ended before 1963,
however, the product in question must be classified as one with a change in
competition regime.

Permanent magnets. A price-fixing agreement by the Permanent Magnet


Association (R999, R1000, R1001, R1935*) was contested in the Court and

3. I report here, for the record, the numbers of all the significant restrictive agreements
between steel manufacturers contained in the register. Brief descriptions of those that
were referred to the Court can be found in the Registrar’s reports. Referred to the
Court: R213*, R245*, R256*, R414*, R864*, R913*, R976*, R1079*, R1081*, R1084*,
R1085*, R1092*, R1093*, R1098*, R1100*, R1101*, R1104*, R1105*, R1106*, R1109*,
R1111*, R1112*, R1113*, R1121*, R1122*, R1123*, R1125*, R1128*, R1129*, R1130*,
R1131*, R1243*, R2445*, R2801*, R2802*, R2803*, R4094*. Not referred to the Court:
R861, R862, R863, R1082, R1083, R1086, R1091, R1099, R1102, R1103, R1107, R1108,
R1110, R1114, R1116, R1117, R1118, R1119, R1120, R1124, R1126, R1127, R1202, R1203,
R2430, R2431, R2432, R2835, R3271, R5069, R5074.
Survey of Collusive Agreements in British Industries 357

was allowed to continue (see LR 3 RP 119, LR 3 RP 392, Swann et al. 1973).


The agreement provided also for joint research, exchange of technical infor-
mation, and patent pooling. This is a secondary product of MLH 311.

Steel Tubes (MLH 312)

An agreement between the two largest producers of steel tubes, Stewards


and Lloyds and Tube Investments, providing for the demarcation of their
respective activities and for common prices for certain classes of tubes that
both firms produced, ended in 1959 (R1187). A similar agreement between
Stewards and Lloyds and a smaller firm was abandoned in 1962 (R1032). A
third agreement of this kind, between Stewards and Lloyds and one of the
largest steel producers, was abandoned in 1959 (R2519).
Moreover, an agreement by the Large Tube Association fixed, until 1963,
the prices of various types of steel tubes and provided for the allocation of
orders received from customers (R677). The Association of Electric Steel
Conduit Manufacturers fixed the prices of steel electric conduits until 1960
(R795, R1161); steel conduits were also listed in CD. The Conduit Fittings
Manufacturers Association set the prices of electrical conduit fittings and ac-
cessories until 1959 (R76). The Cased Tube Association set prices and quotas
for cased tubes until 1957 (R276). Finally, a pricing agreement by a group of
producers of steel pipe fittings ended in 1960 (R1026, R2614). On the other
hand, an agreement by the British Hot Finished Tube Conference, which
ended in 1967, related only to standard conditions for government contracts
(R1029). I should also mention a pricing agreement between two producers
of seamless steel tubes, which came into force in 1973 and was abandoned in
1979 (R5073*, R5078*). In summary, both MLH 312 as a whole and its con-
stituent principal products experienced a change of competition regime as a
result of the 1956 Act.

Iron Castings, etc. (MLH 313)

Pig iron. Several agreements, covering all types of pig iron, were regis-
tered. The Basic Pig Iron Producers Association (R1016, R1097*), the Na-
tional Association of Hematite Pig Iron Makers (R1094*, R1095), the Foundry
Pig Iron Producers Association (R474*) and the Cylinder and Refined Iron
Association (R1087, R1088) all set the prices of their respective products or
recommended that the maximum prices set by the Iron and Steel Board be
treated as fixed prices. Some of these agreements were superseded by others
or ended in the late 1950s. Those that were in force in the early 1960s, cov-
ering more than 90% of total industry sales, were abandoned in 1964–1965.
Thus there was no change in competitive conditions during the period 1954–
1963 in the pig iron industry. After the mid-1960s, the industry was radically
transformed as a result of the nationalization of steel.
358 Appendix A

Marine and engineering iron castings. The National Federation of Engi-


neering and General Ironfounders made recommendations for price
increases in marine and engineering castings to its thirteen member associa-
tions, who then circulated the recommendations to their members (R282,
R2791). In fact, until 1962 several regional associations took a more active
part in price setting, but after 1962 they merely forwarded the recom-
mendations of the federation and, in a few cases, operated information
agreements (see R138, R207, R236, R373, R397, R399, R411, R420, R529, R545,
R587, R731, R1583, R1672). The recommendations ceased after 1965.

Other iron castings. Several price-fixing agreements covered nearly all of


the more specific types of iron castings. In a series of agreements between its
members and with various outside firms, the British Ironfounders Associa-
tion set prices for cast iron rainwater and soil goods, cast iron drain con-
nections, general builders’ castings, and domestic solid-fuel cooking and
heating appliances (R2797*, R2798*, R2799*, R2800*). See MRPC, Report on
the Supply of Cast Iron Rainwater Goods (London: H.M.S.O., 1951) for details of
these arrangements, which initially also involved collective exclusive dealing,
aggregated rebates, and collective resale price maintenance. The registered
agreements were abandoned in 1963. The British Bath Manufacturers Asso-
ciation set minimum prices for cast iron porcelain enameled bathtubs until
1960 (R548*). Note that the bathtub industry is also listed in CD. According
to Swann et al. (1973), the abolition of the agreement was followed by a
prolonged period of very intense competition that lasted throughout the
1960s. The British Malleable Tube Fittings Association set prices for mallea-
ble tube fittings and was also a party to international agreements until 1960
(R661, R664, R1805, R1806, R1807, R1937, R2500). According to BT, it may
also have been engaged in collective discrimination arrangements.
Other price-fixing bodies included the National Ingot Mould Association
(R997), the Cast Iron Pressure Pipe Association (R475), the Range Boiler
Makers Association (R182), the Cast Iron Chair Association (R828), the Cast
Iron Axlebox Association (R893), and the Cast Iron Segment Association
(R918, R919). All these associations abandoned their respective agreements
between 1958 and 1962. According to CD and BT, cast iron heating boilers
and radiators were also subject to collective discrimination arrangements
before 1956. In summary, both the iron casting industry as a whole and
nearly all of its principal products experienced a change in competition re-
gime as a result of the 1956 Act.

Aluminium and Aluminium Alloys (MLH 321)

The register contains restrictive agreements from all three sections of the in-
dustry: smelters, fabricators, and founders. The Federation of Light Metal
Smelters fixed prices for certain types of ingots until 1958 (R192), although
Survey of Collusive Agreements in British Industries 359

these were ‘‘maximum’’ prices, at least according to the registered agree-


ment. The Light Metal Founders Association recommended price increases
to cover changes in costs in the castings industry until 1961 (R419), and
operated an information agreement for some time after that date (Swann et
al. 1974). The British Aluminium Foil Rollers Association fixed prices for
aluminum foil until 1959 (R1074), and operated an information agreement
after that date. Finally, seven leading fabricators were parties to a collu-
sive arrangement for aluminum semimanufactures, which was not initially
registered and therefore continued until 1967 (R3045*, R3046). See also MC,
Aluminium Semi-manufactures (London: H.M.S.O., 1967). The arrangement
provided for discussions about prices, the occasional allocation of large
orders between the parties, and restrictions imposed on distributors for sales
of imported goods. In the long run, then, there was a change of competition
regime for the aluminum industry as a whole and for most, if not all, of its
principal products.

Copper, Brass, and Other Copper Alloys (MLH 322)

Collusive agreements were widespread in this industry, covering its three


sections: smelters, fabricators, and founders. The main data source for
the semimanufactures section of the industry is MRPC, Report on the Supply
and Export of Certain Semi-manufactures of Copper and Copper-based Alloys
(London: H.M.S.O., 1955). This report described the collusive activities of
more than ten associations of fabricators, covering virtually all copper and
copper alloy semimanufactures: plate, sheet, strip, rods, and tubes (and also
wire, which is not part of MLH 322). The associations imposed agreed prices,
aggregated rebates, and, in several cases, collective exclusive dealing. Sev-
eral were also involved in international collusive agreements. Most of the
associations investigated by the MRPC did not register their agreements.
Presumably they abandoned them in 1956, in light of the fact that they had
been severely criticized by the MRPC. Thus the register contains only
agreements by the Rod Rollers Association (R1553), the Cadmium Copper
and Bronze Association (R1554), and the High Conductivity Copper Associ-
ation (R508*). All three ended between 1957 and 1959. It seems that R1554
and R508* were replaced by information agreements for part of the 1960s.
The register also contains one agreement between smelters and another
between founders. The British Bronze and Brass Ingot Manufacturers Asso-
ciation recommended prices of bronze and brass ingots until 1962 (R2732).
After that date, it operated an information agreement, which seems to have
continued until the 1970s. The Merseyside Non-Ferrous Founders Associa-
tion fixed prices for copper and copper alloy castings until 1959 (R586).
Finally, an agreement by the British Non-Ferrous Metals Federation, which
was dropped in 1957, related only to recommended conditions of sale (R410).
In summary, the 1956 Act caused a change in competition regime for MLH
322 as a whole as well as for most, if not all, its principal products.
360 Appendix A

Miscellaneous Nonferrous Metals (MLH 323)

Lead manufactures. The Lead Sheet and Pipe Association specified mini-
mum prices for lead sheet and pipes (R1052*). According to CD, it also en-
gaged in certain forms of collective discrimination, including aggregated
rebates, before 1956. The agreement was formally canceled in 1962. Until that
date the association was also a party to an agreement with two smelters re-
garding maximum buying prices of scrap and remelted lead (R803). More-
over, between 1962 and 1965 several members of the association operated an
effective information agreement (R3075). The register also contains a pricing
arrangement of the Association of Lead Sellers, whose membership consisted
of manufacturers and merchants of lead sheet and pipes in Scotland (R908),
and an arrangement among three smelters for the price of scrap and
remelted lead (R1399). Both of these ended in 1962. Lead sheet and pipes
accounted for 15–20% of total sales of lead manufactures in the 1960s.

Nickel manufactures. No evidence of any restrictive agreements.

Tin manufactures. No evidence of any restrictive agreements.

Zinc manufactures. A price agreement between producers of zinc strip


is mentioned in MRPC, Report on the Supply and Export of Certain Semi-
manufactures of Copper and Copper-based Alloys (London: H.M.S.O., 1955). This
product accounts for less than 10% of total sales of zinc manufactures, how-
ever, so the effect of the 1956 Act on the industry was not significant.

Other products. The Collapsible Tube Association fixed prices for all types
of collapsible tubes made from nonferrous metals until 1958 (R1258). An
agreement between two leading producers of various ferrous alloys, High
Speed Steel Alloys and Murex, relating to prices, allocation of products, and
some profit-sharing (R1374, R2675), ended in 1962.

Agricultural Machinery, Except Tractors (MLH 331)

An agreement by the Agricultural Engineers Association, whose member-


ship consisted of manufacturers and wholesalers of agricultural machinery
and tractors, provided for recommended discounts by manufacturers and
wholesalers to various classes of customers and for conditions of sale (R860).
The agreement was abandoned in 1957. In addition, the manufacturers prac-
ticed resale price maintenance, and there were also, until the late 1960s,
agreements between dealers to maintain the manufacturers’ prices (R789*,
R2846). The industry was also listed in CD and BT for operating collective
discrimination arrangements. The effect of all the above schemes on compe-
tition between manufacturers is rather uncertain, however. There was no
Survey of Collusive Agreements in British Industries 361

agreement on retail prices, which therefore were individually set. Moreover,


each manufacturer appointed a number of distributors and dealers in any
given area (both before and after 1957), and the discounts given by different
manufacturers to their designated distributors and dealers were not regu-
lated by the agreement—in fact, they varied considerably and were the
product of direct negotiations.
The register also contains a series of agreements between two leading
manufacturers of agricultural machinery and tractors, Ford Motor and Ran-
somes Sims and Jeferries, relating to the joint production and/or distribution
of certain types of machinery (R2775, R840, R3197). Some of these agree-
ments were abandoned in the early 1960s, while others continued until the
early 1970s. Finally, the register contains two price agreements, one by the
Agricultural Machine Parts Association (R921) and one by the Milking
Machine Manufacturers Association (R1714). The first of these seems to have
been a regional arrangement. Both were abandoned in the late 1950s. All in
all, the effect of the 1956 Act on the agricultural machinery industry should
probably be regarded as ambiguous.

Metalworking Machine Tools (MLH 332)

Metal-cutting and metal-forming machine tools. The register contains an


agreement involving six machine tool manufacturers who had formed a
‘‘holding company’’ and accepted restrictions as to the types of machine tools
each could produce (R678). Some of them were among the largest firms in
the industry. The arrangement was abandoned in 1959. According to a case
study of the machine tool industry, written by M. E. Beesley and G. W.
Troup and published in Burn (1958), the existence of ‘‘exclusive agencies’’
and the specialization of production within agencies was a common feature
of this section of the industry. Each agency comprised several manufac-
turers, but each manufacturer belonged to a single agency. Without detailed
information on how these agencies operated, their effect on competition can
only be regarded as uncertain. PEP also mentions an agreement by the
Machine Tool Trades Association, which, however, apparently related only
to conditions of sale.

Welding and flame-cutting equipment. No evidence of any restrictive


agreements.

Pumps, Valves, and Compressors (MLH 333)

Pumps. A registered agreement by the British Pump Manufacturers Asso-


ciation, which was abandoned in 1967, related only to conditions of contract
(R425, R1418). However, the association is mentioned in PEP as allegedly
collusive. Moreover, a group of firms in the industry were parties, until 1960,
362 Appendix A

to the ‘‘pump notification scheme’’ (R426). This covered relatively large


pumping equipment and provided for the reporting of inquiries received
from customers, uniform action on terms and conditions with respect to
these inquiries, and the payment by the successful tenderer of a fee to un-
successful tenderers for inquiries that needed a great amount of preparatory
work. However, the scheme did not explicitly provide for price-fixing or the
allocation of work. As I have explained in chapter 3, the effect of such
schemes on competition is not clear.
The register also contains a pricing arrangement among three producers of
semirotary wing pumps (R603*, R1727). This ended in 1959. Two other reg-
istered agreements should probably be classified to MLH 333. The Hydraulic
Association operated a notification scheme for hydraulic plant and also
specified conditions of sale (R232). The agreement was modified in 1966, but
it was not entirely abandoned. Also, a group of three firms were parties,
until 1960, to the ‘‘hydraulic machinery agreement,’’ which covered hydraulic
turbines and associated hydraulic mechanical equipment, and provided not
only for the reporting of inquiries received from customers but also for dis-
cussions about prices (R700). The overall picture for the pump industry is one
of uncertainty: those agreements that were clearly restrictive covered only
certain parts of the industry, while those that clearly covered large sections
of the industry contained restrictions with an uncertain effect on competition.

Valves. No agreements for valves were registered, but the industry was
mentioned in PEP as allegedly collusive.

Compressors. The Portable Air Compressor Association specified mini-


mum prices for portable air compressors until 1959 (R1301*). The agreement
also provided for the exchange of technical information in basic research,
although not in product development. In addition, a group of firms were
parties to the ‘‘centrifugal and axial flow blower and compressor agreement,’’
which regulated prices and tendering fees until 1961 (R717). The combined
share of the products covered by these two agreements in the total sales rev-
enue of the compressor industry was about 50% in the 1960s. Hence the 1956
Act caused a change of competition regime in this industry.

Fluid power equipment. No evidence of any restrictive agreements.

Industrial Engines (MLH 334)

Internal combustion engines. Four agreements, covering various types of


compression ignition engines and providing for the notification of inquiries
and for agreed prices, were abandoned in 1961–1962 (R926, R927, R928,
R929). Also, a price agreement for steam engines (R686) was canceled in 1959.
There was, then, a change of competition regime in this part of MLH 334.
Survey of Collusive Agreements in British Industries 363

Other prime movers. The register contains several agreements relating to


turbines and condensing plant (some of these also covered products classi-
fied to MLH 361). The agreements comprised the notification of inquiries and
orders, prices, provisions regarding the payment of fees by subcontractors to
main contractors, and often design specifications. The ‘‘small turbine agree-
ment’’ (R705), the ‘‘small condenser agreement’’ (R715) and the ‘‘small tur-
bine and small condenser inter-contracting agreement’’ (R1842) ended in
1961. The ‘‘large turbine and large turbo-type alternator inter-contracting
agreement’’ (R1932*) and the ‘‘large turbine and large condenser inter-
contracting agreement’’ (R1933*) were abandoned in 1962. Finally, the ‘‘large
turbine price agreement’’ (R704*) and the ‘‘large condenser price agreement’’
(R685*) were canceled in 1967. Thus this part of MLH 334 also experienced a
change of competition regime as a result of the 1956 Act.

Textile Machinery and Accessories (MLH 335)

The only price agreements for complete machines were the ones registered
by the British Jacquard Engineers’ Association (R515) and the Lace Machine
Builders and Allied Trades Association (R1479). Both were abandoned in
1959–1960. A third agreement, by the British Knitting Machine Builders As-
sociation, comprised only conditions of sale and ended in 1964 (R297). On
the other hand, most parts and accessories of textile machinery were subject
to price fixing until 1959 (R57, R110, R259, R260, R383, R384, R432, R534,
R537, R1427, R1451, R1507). The combined share of jacquard machines and
lace machines in the product category ‘‘complete textile machinery’’ (which
is the one used for the concentration regressions in this book) was probably
around 5%. Even if parts and accessories are taken into account, the collusive
products covered at most 10–15% of total sales revenue of MLH 335. Hence
both the product category ‘‘complete textile machinery’’ and MLH 335 as a
whole can be seen as largely competitive.

Construction and Earth-Moving Equipment (MLH 336)

An agreement by the Excavator Makers Association (R83) provided for the


reporting of inquiries and tenders, maximum discounts to ‘‘resale mer-
chants’’ and the prenotification of price changes. However, neither individ-
ual list prices nor commissions to appointed agents were regulated. The
Concrete Mixers Manufacturers Association specified minimum prices and
commissions to agents and merchants (R958*, R2373*). The members of the
Road Roller Manufacturers Association accepted minimum prices (R880*).
The agreement of the Well Drillers Association comprised the reporting of
inquiries and minimum prices for tenders (R567). Finally, the Rammer
Manufacturers Association set prices for power rammers (R24). All these
agreements were canceled in 1959. The affected products accounted for 20–
30% of total industry sales.
364 Appendix A

Mechanical Handling Equipment (MLH 337)

Conveyors, Aerial Ropeways, etc., and Lifting and Winding Devices. An


agreement by the Mechanical Handling Engineers Association (R225) pro-
vided for the reporting of inquiries received from customers and for tender-
ing fees in connection with a wide range of mechanical handling equipment.
An agreement by the Aerial Ropeways Association (R226) may have been
somewhat more restrictive, since it also contained a reference to ‘‘uniform
action where necessary’’ following the reporting of inquiries; however, it
covered only aerial ropeways. Both arrangements were abandoned in 1965.
A price agreement in reels and winches, which are secondary products of the
industry, was abandoned in 1959 (R868, R1304). Overall, the effect of the
1956 Act on competition in this industry was ambiguous.

Cranes. Until 1963, the agreement of the Association of Crane Makers


(R2578*) provided for the reporting of inquiries and for discussions between
interested members about the terms and conditions for particular tenders
(but not about prices). There were also occasional attempts to allocate work.
After 1963, the agreement comprised the reporting of inquiries, the reporting
of the terms and conditions set by each individual member for any given
inquiry, and the payment of tendering fees. The arrangement was dis-
continued in 1967. In conclusion, while there is some evidence of significant
restrictions on competition in this industry, it is probably not overwhelming.

Lifts and escalators. An agreement by the National Association of Lift


Makers contained recommendations about installation and maintenance
charges, conditions of sale, restrictions on sales promotion, and the require-
ment not to carry out the maintenance of a new lift (elevator) installed by
another member for one year after installation (R1043). Most restrictions
were canceled in 1963. The 1956 Act probably caused a change of competi-
tion regime in this industry.

Powered industrial trucks and industrial tractors. No evidence of any re-


strictive agreements.

Office Machinery (MLH 338)

The register contains an agreement by the Typewriter and Allied Trades


Federation of Great Britain, whose membership consisted of manufacturers,
importers, and distributors of various types of office machinery (R665). The
products covered by the agreement accounted for about 40–50% of total
sales of the office machinery industry. As registered, the agreement provided
for the reporting of individual list prices and trade discounts to the associa-
tion, the maintenance of manufacturers’ individual resale prices, restrictions
Survey of Collusive Agreements in British Industries 365

on sales promotion, and minimum charges for the renting and maintenance
of typewriters and calculating machines in particular. Most of these restric-
tions were canceled in 1957, but those involving the exchange of information
remained. BT also reports allegations of collective discrimination arrange-
ments in typewriters and typewriter components.
Further details on the industry are provided in CD and PEP (in the latter
case the information comes from a manufacturer of typewriters). It seems
that the original agreement was more restrictive than the one registered, at
least for typewriters: it provided for the collective enforcement of resale prices
and may also have regulated individual prices. On the other hand, PEP also
confirms that many products of the industry were not regulated. The effect of
the 1956 Act on the office machinery industry as a whole was therefore
ambiguous. There is also evidence that there was a change of competition
regime for typewriters in particular.

Mining Machinery (MLH 339.1)

An agreement by the Coal Preparation Plant Association provided for


the reporting of inquiries and recommended terms and conditions for con-
tracts with the National Coal Board (R889). This agreement was allowed
to continue without a Court hearing, since it was not regarded as restrictive.
The Skip Plant Association also had a reporting agreement, which made
vague references to uniform action (R1824). The restrictions in question were
dropped in 1960. The products covered by R1824, skip plant and mine car
handling plant, accounted for less than 10% of total industry sales. Two other
agreements by associations in the industry, relating to terms and conditions
for contracts with the National Coal Board, were also allowed to continue
without a Court hearing (R510, R890). In summary, the industry was prob-
ably competitive, a view confirmed by evidence given by the National Coal
Board and reported in PEP.

Printing, Bookbinding, and Paper Goods Making Machinery


(MLH 339.2)

There were no significant registered agreements. BT reports allegations of


collusion in printing blocks, a secondary product of the industry.

Refrigerating Machinery, Except Domestic-type Refrigerators


(MLH 339.3)

An agreement by the British Refrigeration Association (R824), covering both


commercial refrigerating machinery and domestic refrigerators, provided
for the maintenance of the (individually set) manufacturers’ prices and the
exchange of price information between the parties, and also contained max-
366 Appendix A

imum trade discounts, conditions of sale, and restrictions on sales promo-


tion. See also the Lloyds’ report for details. The agreement was abandoned,
possibly in the late 1950s, leading to a change of competition regime in this
industry.

Space-Heating, Ventilating, and Air-Conditioning Equipment


(MLH 339.4)

There is no evidence of significant restrictive agreements in this industry.

Food and Drink Processing Machinery (MLH 339.7)

An agreement by the Mineral Water Engineers Association and the Brewery


Equipment Section of the Food Machinery Association provided for resale
price maintenance, maximum trade discounts, and conditions of sale for
mineral water machinery and brewing machinery (R530). The Dairy Engi-
neers Association had a similar arrangement for dairy machinery (R1540).
The products affected by these agreements, both of which ended in 1958–
1959, accounted for 20–25% of total industry sales.

Miscellaneous Nonelectrical Machinery (MLH 339.5, 6, 8, and 9)

Portable power tools. The Pneumatic Tool Association specified minimum


prices for various types of pneumatic tools until 1959 (R308*). The Pneumatic
Metal Working Tool Association also fixed prices until 1959 (R307). The
products covered by these agreements accounted for about 20–30% of total
industry sales. The industry was also mentioned in PEP as allegedly collusive.

Rolling mills. An agreement by the Steel Works Plant Association (R227)


provided for the reporting of inquiries above a certain value, uniform action
on terms and conditions (but not on prices) with respect to these inquiries,
and tendering fees. Several of the restrictions were abandoned in 1964. The
register also contains an interreporting agreement between the Steel Works
Plant Association and the Hydraulic Association, which was canceled in 1961
(R1359), and an arrangement about commissions between members of the
Steel Works Plant Association and a group of manufacturers of electrical parts
for rolling mills (R698), which ended in 1960. The effect of these agreements
on competition in the rolling mill industry is uncertain.

Lawn mowers. Electric mowers were covered by an agreement of the


Associated Manufacturers of Domestic Electric Appliances, which ended in
1959 (R107). The details are described under MLH 368. Electric mowers
accounted for about 10% of total sales of mowers, so the effect of the 1956
Act on the industry as a whole was probably not significant.
Survey of Collusive Agreements in British Industries 367

Other nonelectrical machinery. An agreement by the Society of Laundry


Engineers and Allied Trades (R1145) provided for resale price maintenance
and conditions of sale, including the requirement not to give discounts to
certain types of buyers, but did not specify prices or trade discounts in gen-
eral. It was abandoned in 1965. An agreement by the Garage Equipment
Association, whose membership consisted of both manufacturers and dis-
tributors, comprised arrangements for collective exclusive dealing, resale
price maintenance, agreed discounts, and conditions of sale (R793*); the in-
dustry was also listed in CD. The agreement was abandoned in 1960, and
was replaced by an information agreement, which remained in force until
1972. BT also reports allegations of collusion in fire sprinkler equipment.

Industrial (Including Process) Plant and Steelwork (MLH 341)

Boilers and boilerhouse plant. An agreement by the Water-Tube Boiler-


makers Association comprised a procedure for tendering that effectively
resulted in the sharing of orders between member firms (R509*). It was re-
ferred to the Court, contested by the parties, and allowed to continue (LR 1
RP 285). Several other agreements, relating to conditions of contract with the
Central Electricity Generating Board (the main buyer of water-tube boilers),
remained in force throughout the 1960s and early 1970s (R1593, R3728,
R1486, R3121). On the other hand, a price-fixing agreement by the Associa-
tion of Shell Boilermakers was abandoned in 1957 (R391). Although water-
tube boilers were the main product of the industry, shell boilers accounted
for about 20% of total industry sales in the late 1960s.

Other industrial plant, including process plant. An agreement by the


Gasholder Makers Association provided for the reporting of inquiries and
minimum prices for larger types of gasholders and gasholder tanks (R1073).
The prices of furnaces and kilns were fixed in accordance with the ‘‘electric
resistance furnace agreement’’ (R818*) and the ‘‘electric melting furnace
agreement’’ (R819), both of which also provided for the reporting of inquiries.
All three agreements were abandoned between 1958 and 1960. The affected
products accounted for more than 20% but less than 50% of total industry
sales.

Constructional steelwork and ironwork. A price-fixing agreement by the


British Constructional Steelwork Association (R210*, R229) was abandoned
in 1959. A separate agreement, providing for the reporting of inquiries and
orders and specifying conditions of sale, continued throughout the 1960s and
the 1970s (R3151). An agreement by the Bridge and Constructional Ironwork
Association, which contained recommendations about price increases to
cover changes in costs (R228), was canceled in 1959, and an interreporting
arrangement between the two associations ended in 1960 (R1600). Hence
there was a change of competition regime in this industry.
368 Appendix A

Other fabricated ironwork and steelwork. An agreement by the Tank and


Industrial Plant Association in connection with storage tanks and pressure
vessels provided for the reporting of inquiries received and orders placed,
the payment of tendering fees, and, in some cases, the allocation of work
(R304). The last of these provisions was withdrawn in 1960, and the payment
of tendering fees was discontinued in 1965. Additional information from PEP
(provided, in this case, by a firm in the industry) suggests that the associa-
tion’s reporting procedure was designed to achieve ‘‘fair prices’’ and to resist
‘‘auctioning’’ by buyers. Hence it is probably fair to say that the 1956 Act
caused a change of competition regime in this industry.

Ordnance and Small Arms (MLH 342)

There is no evidence of any significant restrictive agreements in this industry.

Other Mechanical Engineering Not Elsewhere Specified (MLH 349)

Ball and roller bearings. No agreement was registered, but both PEP and
BT mention this industry as allegedly collusive.

Rolls of iron and steel. The National Forgemasters Association recom-


mended prices for forged rolls until 1962 (R1122*). Also, until 1962, the Roll
Makers Association fixed the prices of cast iron and steel rolls (R476). Infor-
mal discussions and occasional agreements on prices between some of the
leading firms in the cast roll section of the industry continued until the late
1970s (R5075). There is some uncertainty about the effectiveness of these
arrangements, however, especially after the nationalization of the steel in-
dustry in 1967 (which created a dominant buyer for cast rolls). Thus, although
the iron and steel roll industry was collusive in the 1950s, the effect of the 1956
Act on the industry during the period examined in this book must be regarded
as ambiguous.

Other products. Producers of gears reported inquiries and fixed prices


in accordance with the ‘‘turbine reduction gear agreement’’ (R692) and the
‘‘heavy reduction gear agreement’’ (R706) until 1960–1961. The Railway Cast
Bearings Association allocated orders among its members until 1958 (R1030);
this agreement covered only a small part of the bearing industry, however. A
price-fixing agreement by the Tuyere Makers Association (R91*) was aban-
doned in 1959. An agreement by the British Oil Burner Manufacturers As-
sociation, relating only to conditions of tender and trading, ended in 1964
(R1144). Piston rings may have been subject to collective exclusive dealing,
according to BT. Finally, four other price-fixing agreements in the register
should probably be classified under MLH 349: R53, between producers of
gas governors; R683, by the Gedges Drawback Hook Manufacturers Associ-
ation; R920, by the Buffer Rod Makers Association; and R1530*, by the
Survey of Collusive Agreements in British Industries 369

Pressed Bowl Makers Association. The first three of these ended in 1959, and
R1530* was canceled in 1964.

Photographic and Document Copying Equipment (MLH 351)

There is no evidence of collusion in the 1950s. However, a number of price


agreements relating to diazo copying machines and drawing office equip-
ment were discovered by the MMC in the course of its inquiry into the diazo
copying materials industry (R4077*, R4084*, R4085*). See MMC, Diazo Copy-
ing Materials (H.M.S.O., 1977) for details. These agreements were apparently
in operation for a limited period, from 1969 to 1972, so the industry can
probably be classified as one without a change in competition regime be-
tween the 1950s and the 1970s.

Watches and Clocks (MLH 352)

Watches. No evidence of any significant restrictive agreements.

Clocks. The British Impulse Clock Manufacturers Association recom-


mended prices for various types of electrical impulse clocks until 1959 (R745).
PEP also reports allegations of a collusive agreement between members of
the British Synchronous Clock Conference (which may be the predecessor of
B.I.C.M.A.) The share of electric clocks in the total sales of MLH 352 was
about 20%. According to CD, the mechanical clock industry was subject to
collective discrimination arrangements before 1956.

Surgical Instruments and Appliances (MLH 353)

Medical, surgical, dental, and veterinary instruments and appliances.


The register contains several agreements on surgical appliances. The Surgical
Appliance Manufacturers Association set minimum prices for surgical sup-
ports and surgical elastic hosiery until 1963 (R2827). The British Surgical
Trades Association recommended price increases for a wide range of surgi-
cal appliances, including elastic hosiery and surgical footwear, until 1963
(R2816*). The Scottish Surgical Instrument Manufacturers Association also
recommended prices for surgical appliances, including elastic hosiery and
surgical footwear, until 1967 (R3160*). A group of manufacturers of balloon
catheters fixed prices until 1963 (R2751*). Finally, the Chemists Sundries
Association fixed the prices of various pharmacists’ and hospital sundries
manufactured in earthenware until 1959 (R955). The share of the products
covered by these agreements in total industry sales was about 30%. I should
also mention an agreement by the Association of Dental Manufacturers and
Traders of the United Kingdom, which provided for collective exclusive
dealing and the collective enforcement of resale price maintenance, but did
370 Appendix A

not regulate the manufacturers’ prices or trade discounts. See the Lloyds’
report and MRPC, Report on the Supply of Dental Goods (London: H.M.S.O.,
1950) for details.

Spectacles and lenses. Prices for spectacles, frames, and lenses were set by
four different associations: the Ophthalmic Prescription Manufacturers As-
sociation (R672), the British Metal Spectacle Manufacturers Association
(R673), the Plastic Spectacle Manufacturers Association (R674), and the Brit-
ish Ophthalmic Lens Manufacturers Association (R675). All four agreements
were abandoned in 1959–1960, leading to a change in competition regime in
this industry.

Scientific and Industrial Instruments and Systems (MLH 354)

Optical instruments. No evidence of any restrictive agreements.

Other scientific and industrial instruments and systems. The Meter


Manufacturers Association set minimum prices for electricity meters (R483,
R484). This agreement was formally abandoned in 1960 and was replaced by
an information agreement (R3161*), which apparently was successful in
reducing competition in the industry until 1970. See also MMC, Electricity
Supply Meters (London: H.M.S.O., 1979). The Summation Meter Manufac-
turers Association set minimum prices for summation meters until 1965 (R878*,
R1492*). The prices of various other types of instruments for measuring
electrical magnitudes were regulated by the ‘‘subsidiary power instrument
agreement’’ (R694) and the ‘‘patterns group agreement’’ (R691) until 1958–
1960. In addition, the Commercial Instruments Conference fixed the prices of
certain types of ammeters and voltmeters until 1961 (R708). Finally, the pri-
ces of gas meters were set, until 1957, by the Gas Meter Makers Conference
(R669). The share of all the products affected by these agreements in total
sales of scientific and industrial instruments and systems (excluding optical
instruments) was about 10%. This was, then, a largely competitive industry.

Electrical Machinery (MLH 361)

Generators and turbines. The agreements in generators and turbines com-


prised prices, the reporting of inquiries and orders, the payment of fees
by subcontractors to main contractors, and sometimes design specifications
or exchange of technical information. Before 1956 they also provided for
certain forms of collective discrimination, such as aggregated rebates. The
‘‘large turbo-type alternator price agreement’’ (R688*) was canceled in 1967.
Five other agreements—the ‘‘generator price agreement’’ (R702), the ‘‘small
turbine-driven alternator and generator agreement’’ (R716), the ‘‘small tur-
bine and turbine-driven alternator and generator inter-contracting agree-
ment’’ (R1843), the ‘‘alternator price agreement’’ (R713), and the ‘‘marine
Survey of Collusive Agreements in British Industries 371

turbo-generator agreement’’ (R703)—ended in 1960–1961. In addition, many


of the agreements reviewed under MLH 334 are also relevant here (espe-
cially R705, R1842, R1932*, R1933*, and R704*). An arrangement between the
Steam Turbine and Associated Plant Manufacturers and the Central Elec-
tricity Authority (the largest buyer of electrical machinery), regarding dis-
counts for various types of machinery, ended in 1967 (R728). Finally, the
‘‘gas turbine generating plant agreement’’ (R2793) was canceled in 1969.
Further details on the operation of these and other agreements in the elec-
trical machinery industry, including some that involved only the reporting
of inquiries, can be found in MRPC, Report on the Supply and Exports of Elec-
trical and Allied Machinery and Plant (London: H.M.S.O., 1957). In conclusion,
this part of MLH 361 experienced a change of competition regime as a result
of the 1956 Act.

Transformers. The principal agreement for transformers and related


equipment was operated by the Associated Transformer Manufacturers and
provided for minimum prices, the reporting of inquiries and orders, and
aggregated rebates (R707*). It was referred to the Court, defended by the
parties, and struck down in 1961 (LR 2 RP 295, LR 7 RP 202). A second
agreement, covering smaller transformers, was abandoned in 1958 (R690), as
was an ancillary arrangement involving one firm which was not a member of
A.T.M. (R2245). See MRPC, Report on the Supply and Exports of Electrical and
Allied Machinery and Plant (London: H.M.S.O., 1957) for details on these
agreements, and Swann et al. (1973) for an account of the emergence of
competition in the industry following the collapse of an information agree-
ment in the mid-1960s. The register also contains a number of annual agree-
ments/contracts between manufacturers of large transformers and the
Central Electricity Generating Board (the successor to the Central Electricity
Authority), relating to the Board’s requirements of very large transformers
during the period 1970–1975 (R3616, R3700, R3746). All in all, there was a
change of competition regime in this industry.

Switchgear. The register contains two agreements that covered all but the
smaller sizes of switchgear (R689*, R711*). They comprised prices, the report-
ing of inquiries and orders, and design specifications. They were canceled
in 1965–1967. A third agreement, which related to mining-type switchgear,
ended in 1959 (R722). On the other hand, smaller switchgear was subject to
an agreement that involved the reporting of inquiries and conditions of sale,
but did not regulate prices (R695); the arrangement was nevertheless can-
celed in 1960. The register also contains several agreements between manu-
facturers and the Central Electricity Generating Board, regarding prices to be
paid for particular orders of heavy switchgear (R2904, R3077, R3673, R3693).
These continued until the mid-1970s, but it is doubtful whether they restricted
competition to any significant extent. In any case, heavy switchgear ac-
counted for a relatively small part of total industry sales in the 1970s. Thus
the 1956 Act caused a change of competition regime in this industry.
372 Appendix A

Motors. Several restrictive agreements, covering various types of motors,


were registered and then abandoned between 1958 and 1965: the ‘‘large elec-
tric machine agreement’’ (R697), the ‘‘fractional horsepower motor agree-
ment’’ (R693*), the ‘‘dynamo and motor agreement’’ (R714), and the ‘‘marine
motor and generator agreement’’ (R709). R697 and R709 also covered gen-
erators. Some of these were replaced for some time by information agree-
ments. See MRPC, Report on the Supply and Exports of Electrical and Allied
Machinery and Plant (London: H.M.S.O., 1957) for details on the collusive
agreements of the 1950s, and Swann et al. (1973) for a description of the
emergence of effective competition in the industry in the 1960s. There was a
change of competition regime in the electric motor industry.

Other electrical machinery. Nearly all other types of electrical machinery


were subject to collusion: control gear for motors (R701, R720), small switch
and fuse gear (R668), electrical parts for rolling mills (R712), electrical
equipment for winding or haulage engines (R699), electrical power distribu-
tion equipment (R658), railway traction electrical equipment (R2120), elec-
trical equipment for diesel/electric main line locomotives (R2121), mercury
arc rectifiers (R687), and protective transformers, relays, and accessories
(R696). All these agreements ended between 1959 and 1961. Finally, an
agreement on installation charges between the electrical machinery sections
of the British Electrical and Allied Manufacturers Association and the Cen-
tral Electricity Authority and the Area Electricity Boards ended in 1974.

Insulated Wires and Cables (MLH 362)

Collusive agreements were widespread in this industry, covering all types of


electric cables. In fact, few industries were as tightly regulated as this one. The
agreements provided for common prices, quotas, collective exclusive dealing,
aggregated rebates, collective resale price maintenance, and patent pooling.
There were also international agreements limiting imports. See MRPC, Report
on the Supply of Insulated Electric Wires and Cables (London: H.M.S.O., 1952).
The agreements were registered, although some restrictions—such as quotas,
aggregated rebates, and collective exclusive dealing—were dropped after
the publication of the MRPC report or before registration. In particular, the
register contains agreements by the Mains Cable Manufacturers Association
(R1462), the Mains Cable Manufacturers Association (Super Tension) (R1461*,
R2859*), the Telephone Cable Makers Association (R798, R1555, R1556,
R1685), the Rubber and Thermoplastic Cable Manufacturers Association
(R1541), the Covered Conductors Association (R1523), the Independent Cable
Makers Association (R750), the Association of Plastic Cable Makers (R1042),
and the Switchboard Cable Association (R1557). Most of the agreements were
abandoned between 1959 and 1961, except for those on supertension mains
cables and accessories, which were finally canceled in 1969. Supertension
mains cables accounted for about 15% of total sales of mains cables (and less
than 10% of sales of all cables) in the late 1960s.
Survey of Collusive Agreements in British Industries 373

Swann et al. (1973) describe the evolution of competition in the industry


during the 1960s. Despite parallel pricing and information agreements, com-
petition emerged first for general wiring cables, then for mains cables, and
finally, after 1969, for supertension mains cables. On the other hand, Swann
et al. thought that the information agreement in telephone cables was rela-
tively more effective in reducing competition. An official investigation into
the industry in the 1970s reached similar conclusions, and in fact revealed a
number of collusive arrangements between producers of telephone cables,
which had been in operation since the 1960s—although they are unlikely to
have had an effect comparable to that of the tight cartels of the 1950s. See
MMC, Insulated Electric Wires and Cables (London: H.M.S.O., 1979). These
agreements (R3850*, R3851*, R3852*, R3853*) were abandoned in 1974–1975.
In summary, there was a change of competition regime for the industry as a
whole and also for most types of cables. The only exception was the tele-
phone cable section of the industry, where competition did not fully emerge
during the period examined in this book.

Telegraph and Telephone Apparatus and Equipment (MLH 363)

The evolution of competition in this industry is summarized in Hart et al.


(1973). The dominant buyer throughout the period examined in this book
was the Post Office. Until 1969, sales of telephone installations to the Post
Office were regulated by an agreement between manufacturers and the Post
Office, which provided for the sharing of orders between the firms and
specified the prices to be paid and conditions of sale. An ancillary arrange-
ment provided for the exchange of technical information and patent pooling
(R3376). Telephone installations are the main product of MLH 363. Another
principal product is line apparatus for long-distance communication. Sales of
line apparatus to the Post Office were similarly regulated, but only until
1963. After 1963 the Post Office gradually introduced more competition in
the tendering process. These agreements were not registered, because the
Crown (and hence the Post Office) was not bound by the provisions of the
1956 Act (see LR 3 RP 98, LR 3 RP 462). On the other hand, the register con-
tains several agreements relating to sales to other buyers or agreements to
which the Post Office was not formally a party: R406, R807, R954, R1135,
R1136, R1452, R2210, R2249, R3813, R3961, R4036, R4097. The register also
contains agreements to which the Post Office was a party and which were
made from the mid-1960s onward (R3377, R3378, R3721). Rather than
describing each particular agreement, I will provide a summary of the evi-
dence contained in the register.
First, the evidence confirms that collusion continued well into the 1960s
with respect to sales of telephone installations to the Post Office (R3377,
R3378). Second, all agreements relating to sales to other buyers were formally
abandoned between 1958 and 1960. Third, an agreement between two leading
firms on line apparatus (R2249) was canceled in 1965 (i.e., soon after the Post
Office introduced competitive tendering for that product). Finally, R3721
374 Appendix A

was a restrictive agreement between manufacturers and the Post Office,


made in the early 1970s and relating to sales of telephone installations to the
Post Office. This suggests that competition emerged slowly in this section of
the industry after 1969, a point also made by Hart et al. (1973). In conclusion,
one can perhaps distinguish two main stages in the evolution of competition
in this industry between the 1950s and the 1970s. Before 1963, both MLH 363
as a whole and all its principal products were collusive and did not experi-
ence any significant change in competitive conditions—given that the bulk
of industry output was sold to the Post Office. After 1963, and until at least
the mid-1970s, the state of competition in MLH 363 as a whole and in tele-
phone installations in particular can only be regarded as ambiguous. How-
ever, it is probably fair to say that there was a change of competition regime
for line apparatus.

Radio and Electronic Components (MLH 364)

Electronic valves, cathode ray tubes, and other active components. The
main association in this industry was the British Radio Valve Manufac-
turers Association, which consisted of producers of electronic valves and
cathode ray tubes for broadcasting or TV receivers. The association members
agreed on common prices for ‘‘maintenance valves’’ (which accounted for
about 10–15% of the total market for valves in the late 1950s) and common
discounts for sales of ‘‘equipment valves’’ to smaller buyers (which accounted
for about 20% of total sales of equipment valves). They also practiced col-
lective exclusive dealing, enforced resale prices, and imposed some restric-
tions on imports and the introduction of new products. See the Lloyds’
report, CD, and especially MRPC, Report on the Supply of Electronic Valves and
Cathode Ray Tubes (London: H.M.S.O., 1957). Following the introduction of
the 1956 Act, the restrictions on common prices and the collective enforcement
of resale price maintenance were withdrawn, and the rest of the agreement
was registered (R313*). It was abandoned in 1959. A pricing agreement by
the Association of Tube Rebuilders, relating to rebuilt cathode ray tubes, also
ended in 1959. The share of products covered by the pricing restrictions
(other than resale price maintenance) in total industry sales revenue was
probably about 20%.

Passive radio and electronic components. No evidence of restrictive


agreements.

Broadcast Receiving and Sound Reproducing Equipment (MLH 365)

Gramophone records and tape recordings. BT reports allegations of col-


lusive pricing and collective discrimination arrangements in this industry,
although the only registered agreements date from the late 1960s and relate
mostly to sales promotion. However, there are some vague references in one
Survey of Collusive Agreements in British Industries 375

of these agreements to past restrictive practices in the industry, although


neither the nature nor the coverage of these is very clear (R3306, R3650). The
effect of the 1956 Act on competition in this industry was ambiguous.

Other broadcast receiving and sound reproducing equipment. The agree-


ment of the Typewriter and Allied Trades Federation of Great Britain (R665),
which was described under MLH 338, covered dictating machines as well.
In addition to collective resale price maintenance and the reporting of indi-
vidual prices and trade discounts, R665 provided for minimum prices for the
renting and maintenance of dictating machines, and maximum discounts
for sales of dictating machines to radio and TV retailers. Most of these
restrictions were canceled in 1957, although those involving the exchange
of information remained. Sales of dictating machines were only a small part
of total sales of broadcast receiving and sound reproducing equipment. The
register also contains an agreement by the British Radio Equipment Manu-
facturers Association, relating to the use of components produced outside
the UK and to participation in exhibitions (R2115*). This ended in 1961, but
its restrictions were not very significant. The industry as a whole was
competitive.

Electronic Computers (MLH 366)

There is no evidence of any restrictive agreements in this industry.

Radio, Radar, and Electronic Capital Goods (MLH 367)

An agreement by the Electro-medical Trade Association, which comprised


the requirement not to give discounts to buyers of certain types of electro-
medical and allied apparatus, was abandoned in 1959 (R117). A pricing ar-
rangement between producers of flameproof electrically operated gate end
boxes ended in 1958 (R721). An agreement by two firms regarding technical
collaboration and market sharing in vehicle-actuated traffic signal equipment
was canceled in 1968 (R145). Finally, an agreement among three manufac-
turers, the Central Electricity Authority, and the South of Scotland Electricity
Board for the manufacture and installation of indicating equipment, speci-
fying the division of business among the three firms (R37, R684), expired in
1969. The combined share of these products in total industry sales revenue
was less than 10%. Hence the effect of the 1956 Act on MLH 367 as a whole
was not significant.

Electric Appliances Primarily for Domestic Use (MLH 368)

An agreement by the Associated Manufacturers of Domestic Electric Cookers


(R106*), covering electric cookers (stoves) and accessories, comprised agreed
increases to individual list prices, maximum discounts to distributors, the
376 Appendix A

maintenance of individual retail prices, the prenotification of any changes in


individual prices, and provisions regarding sales promotion. It was aban-
doned in 1959. An agreement by the Associated Manufacturers of Domestic
Electric Appliances (R107)—covering washing machines, electric heaters,
vacuum cleaners, electric lawn mowers, and several smaller appliances—
provided for maximum discounts to various classes of distributors, the
maintenance of individual list prices, and restrictions on sales promotion.
The agreement ended in 1959. The Associated Manufacturers of Domestic
Electric Water Heaters fixed prices until 1957 (R258). Domestic refrigerators
were covered by the agreement of the British Refrigeration Association
(R824), described under MLH 339.3, and were also listed in CD. The register
also contains price agreements for electric ceiling fans (R718) and desk and
bracket-type electric fans (R719), both of which ended in 1960. Finally, BT
reports allegations of collective discrimination arrangements in electric
appliances. All in all, there was a change of competition regime in the in-
dustry as a whole, as well as for most types of electric appliances (especially
larger appliances). More details on the evolution of competition in washing
machines and refrigerators have been given in chapter 5, section 5.6.

Miscellaneous Electrical Goods (MLH 369)

Electrical equipment for motor vehicles, cycles, and aircraft, excluding


accumulators. No agreements were registered in this industry. According
to MC, Report on the Supply of Electrical Equipment for Mechanically Propelled
Land Vehicles (London: H.M.S.O., 1963), there was, until 1956, an agreement
between two of the largest firms in the industry, Smith and Lucas, which
provided for the demarcation of the firms’ respective activities. It dated from
the 1930s and also covered products in other engineering industries. How-
ever, it was probably of little significance in the 1950s, since the two firms
had by then established their respective products. The MC report also
described various pricing arrangements between manufacturers of spark
plugs, in addition to the collective enforcement of individual retail prices
(which ended in 1956). These arrangements apparently covered only a frac-
tion of the total sales of spark plugs. This product was also listed in BT as
allegedly collusive, but no information was given on coverage. In any case,
spark plugs accounted for less than 10% of the total market for electrical
equipment for motor vehicles, cycles, and aircraft. Hence the 1956 Act had
no significant effect on the industry as a whole.

Primary batteries. No agreements were registered, but the industry was


reported in PEP as allegedly collusive, and was also listed in CD and BT as
being subject to collective discrimination arrangements.

Secondary batteries (accumulators). The main sources of information for


the evolution of competition in this industry are MC, Report on the Supply of
Survey of Collusive Agreements in British Industries 377

Electrical Equipment for Mechanically Propelled Land Vehicles (London: H.M.S.O.,


1963) and Swann et al. (1973). The British Starter Battery Association fixed
prices for automotive replacement batteries, practiced collective exclusive
dealing, and generally regulated the relations between manufacturers and
buyers; the industry was also listed in CD. A modified version of this
agreement was registered (R655), then abandoned in 1960. It was, however,
replaced by an information agreement, which lasted until the mid-1960s. The
MC report also mentions an understanding between the two leading manu-
facturers, Lucas and Chloride, relating to prices and market shares in the
automotive initial equipment market, and an informal pricing arrangement
on traction batteries, both of which presumably were terminated in 1956 and
were never registered. On the other hand, the register contains an agreement
between Chloride and a large manufacturer of primary batteries relating to
the demarcation of their respective activities and to prices for selling one
another’s products (R3743). It also contains several two-party agreements
between producers of train, marine, and commercial vehicle batteries, relat-
ing to terms for selling one another’s products (R3744, R3789, R3790). These
arrangements were revised several times before they were finally canceled
in the 1970s. All in all, there was a change of competition regime in the sec-
ondary battery industry.

Electric lamps. This industry was tightly regulated in the 1950s. The
agreement of the Electric Lamp Manufacturers Association provided for
common prices, quotas, collective exclusive dealing, aggregated rebates, the
collective enforcement of resale price maintenance, patent pooling, and
restrictions on granting licenses to nonmember firms. There were also inter-
national agreements limiting imports. See the Lloyds’ report, CD, and espe-
cially MRPC, Report on the Supply of Electric Lamps (London: H.M.S.O., 1951).
Some of these restrictions were dropped either as a result of the MRPC in-
quiry or before registration, and so the registered agreement contained only
common trade discounts and the exchange of information on individual
prices (R1548, R2487). These were formally abandoned in 1959. However,
various arrangements in the industry continued throughout the 1960s,
including a price information agreement which led to tacit understandings
about changes in prices and discounts (R3725) and the prenotification of any
new product introductions (R3134, R3283). The industry was officially
investigated for a second time in the late 1960s, and the report concluded
that although competition had increased somewhat since 1956, it was still
not very effective. See MC, Electric Lamps (London: H.M.S.O., 1968). Follow-
ing the investigation, the restrictive arrangements apparently ended. In con-
clusion, the emergence of competition in the electric lamp industry was
gradual and slow, but by the early 1970s the industry had probably experi-
enced a change of competition regime.

Electric light fittings, wiring accessories, etc. A pricing agreement by the


Electric Light Fittings Association was abandoned in 1960 (R579*, R1718).
378 Appendix A

According to CD, the association also practiced collective exclusive dealing


before 1956. An agreement by the Association of Manufacturers of Electric
Wiring Accessories, relating to the maintenance of individual list prices and
common trade discounts and conditions of sale, was canceled in 1959 (R667).
Finally, an agreement by the Electric Sign Manufacturers Association, con-
taining some pricing restrictions, was abandoned in 1962 (R1620). In sum-
mary, the 1956 Act caused a change of competition regime in this industry. 4

Shipbuilding and Marine Engineering (MLH 370)

Shipbuilding. No evidence of restrictive agreements.

Boatbuilding. The register contains agreements by the Ship and Boat


Builders’ National Federation (R2854*) and the Marine Traders Association
(R2792*), covering boats and marine equipment. The agreements specified
whether trade discounts should or should not be given to particular classes
of buyers, but did not regulate the discount rates. Although they were both
abandoned in 1964, they were not particularly restrictive. A third agreement,
by the Fishing Boats Builders Association (R2311), provided for minimum
prices, but it covered only Scotland; it ended in 1963. On the basis of this
information, it seems that the bulk of the industry was not subject to signif-
icant restrictive agreements in the 1950s.

Marine engineering. No evidence of restrictive agreements.

Ship repairing. A price-fixing agreement by the North East Coast Ship


Repairers Association seems to have continued throughout the 1960s (R1597).
An agreement by the Dry Dock Owners and Repairers Central Council was
mainly about terms and conditions of contract, but it also contained, until
1966, some minor price restrictions (R1416). The state of competition in this
industry should probably be regarded as ambiguous.

Wheeled Tractors (MLH 380)

Tractors were covered by several of the agreements described under MLH


331: R860, by the Agricultural Engineers Association, and also R2775, R840,
and R3197. Moreover, resale price maintenance was practiced in this in-
dustry. The effect of the 1956 Act on the tractor industry must be regarded as
ambiguous.

4. I should also mention an agreement among several electrical engineering associa-


tions that formed the British Electrical Industry Fair Trading Council (R788). The
agreement specified maximum discounts for various classes of buyers, and was ter-
minated in 1960. It covered a wide range of products, mostly from MLH 368 and MLH
369. See also the Lloyds’ report for details.
Survey of Collusive Agreements in British Industries 379

Motor Vehicles (MLH 381)

Cars and commercial vehicles. An agreement by manufacturers and im-


porters of cars and commercial vehicles was referred to the Court in 1960,
and was unsuccessfully contested by the parties (R1811*; see also LR 2 RP
173). Its main restrictions were the maintenance of individual resale prices
and specified discounts to fleet users. However, neither individual prices nor
discounts to buyers other than fleet users, including discounts by each man-
ufacturer to his franchised dealers, were regulated. According to the Court
report, about 5–10% of all cars and 25–40% of all commercial vehicles were
sold to fleet users. The register also contains an agreement by the Society of
Motor Manufacturers and Traders relating to leasing terms and credit sale
terms for motor vehicles (R1077); this also ended in 1960. Before 1956, the
main organization of the industry, the British Motor Trade Association, col-
lectively enforced the resale prices of cars, commercial vehicles, and motor
accessories, and imposed barriers to entry into distribution; however, it did
not influence individual prices or trade discounts. Details can be found in the
Lloyds’ report, CD, and PEP. On the basis of this evidence, the car industry
can be regarded as largely competitive, while the commercial vehicle industry
should probably be seen as ambiguous.

Motor bodies, engines, and other parts and accessories of motor vehicles.
The only agreement in the register is a price agreement by the Bus Seat
Frame Association (R957), which was abandoned in 1959. Bus seats are a
very minor product of this industry, which was therefore essentially com-
petitive. Parts and accessories of motor vehicles were also covered by the
collective discrimination arrangements operated by the British Motor Trade
Association; these did not influence the manufacturers’ individual prices or
trade discounts.

Trailers and caravans. The register contains an agreement by the National


Caravan Council, providing for fixed trade discounts, the maintenance of
individual resale prices, conditions of sale, and collective exclusive dealing
(R1075*). It was abandoned in 1960. Caravans accounted for about 40–60%
of total sales revenue in the trailer and caravan industry between the mid-
1950s and the early 1970s. Moreover, producers of caravans were generally
fewer and of larger average size than producers of trailers. This implies that
the concentration ratio for the industry as a whole should be influenced more
by the caravan section than by the trailer section, while the reverse should be
the case for the number of firms.

Motorcycles, Tricycles, and Pedal Cycles (MLH 382)

An agreement by the British Cycle and Motor Cycle Industries Association,


whose membership consisted of manufacturers and importers, provided for
common discounts to different types of distributors, the maintenance of the
380 Appendix A

firms’ individual retail prices, the prenotification of any changes in individ-


ual prices or any new model introductions, conditions of sale, and various
restrictions on sales promotion (R531, R533, R2457*, R2789*). Many of these
restrictions covered both bicycles and motorcycles (and accessories), but
some of the most important ones (such as the fixing of trade discounts) ap-
parently covered only motorcycles. Bicycles were also the subject of an
agreement by the Cycle Trade Union, relating to resale price maintenance,
sales promotion, and conditions of sale (R532*). Additional information on
these arrangements prior to 1956 can be found in CD and in the Lloyds’ re-
port. Until 1956, these associations also practiced various forms of collective
discrimination, including the collective enforcement of resale price mainte-
nance and collective exclusive dealing. All significant restrictive arrange-
ments ended between 1962 and 1964.
The register also contains an agreement between the two leading bicycle
manufacturers, Tube Investments and Raleigh, which related, among other
things, to the demarcation of their respective activities with respect to bicycle
tubing and certain bicycle accessories, and to prices charged for these prod-
ucts to one another and to other manufacturers of bicycles (R1184). This
agreement ended in 1960, but its implications for competition in bicycles (as
opposed to bicycle parts and accessories) are not clear. In conclusion, there is
sufficient evidence of a change in competition regime for motorcycles. The
situation for bicycles is somewhat more uncertain. The share of bicycle sales
in total sales of the bicycle and motorcycle industry was about 50–60% for
most of the relevant period.

Aerospace Equipment Manufacturing and Repairing (MLH 383)

The main source of information on this industry is the Report of the Committee
of Inquiry into the Aircraft Industry (London: H.M.S.O., 1965). According to
this report, competition was not very effective, especially for aircraft and
airframes, partly because of the heavy government involvement in the in-
dustry. In particular, competition in aircraft and airframes was being
reduced by government procurement procedures that reflected a wish to
share the available work among the firms. Competition in aero-engines was
more pronounced, at least in the design stage. In the later stages of projects,
competition was affected by the existence of production-sharing arrange-
ments, which the government had approved. These were not registered; in
fact, the only agreement in the register relates to propellers (R41), which is a
minor product of MLH 383. This agreement ended in 1959. On the whole, the
1956 Act had little effect on this industry. Nevertheless, aircraft in particular
was not included in any of the samples used in this book.

Locomotives and Railway Track Equipment (MLH 384)

The register contains several agreements between private manufacturers of


locomotives, all of which provided for the reporting of inquiries and orders
Survey of Collusive Agreements in British Industries 381

and/or specified terms and conditions of contract: R1455, R1827, R1828,


R1829, R2894 and R2895. These were agreements by various sections of the
Locomotive and Allied Manufacturers Association, and were revised or
abandoned between 1961 and 1967. This industry was excluded from my
samples because government establishments accounted for more than two-
thirds of total output during the 1950s and the 1960s.

Railway Carriages, Wagons and Trams (MLH 385)

An agreement by the Railway Carriage and Wagon Building Association


contained prices, the allocation of work, and terms and conditions of con-
tract (R1558, R2671). It was abandoned in 1962. This industry, too, was ex-
cluded from my samples because government establishments accounted for
more than two-thirds of total output during the 1950s and the 1960s.

Engineers’ Small Tools and Gauges (MLH 390)

The register contains price-fixing agreements by the Twist Drill Association


(R478*), the Milling Cutter and Reamer Association (R479), the Flexible Back
Bandsaw Manufacturers Association (R413), the Tubular Frame Saw Associ-
ation (R904), the Short Saw and Crosscut Saw Association (R948), the Circular
and Long Saw Association (R977), the British Hacksaw Makers Association
(R1497, R1670), the British Hard Metal Association (R1041), the High Speed
Steel Tool Bit Association (R1080*), the High Speed Steel Drill Rod Associa-
tion (R1115*, R2488*), the Welded Tool Manufacturers Association (R417),
the Screw Thread Tool Manufacturers Association (R481), and by a group of
producers of thread gauges (R896). Several of these agreements will also
be listed under MLH 391, since they seem to have covered products in both
industries. All the agreements were abandoned in 1959–1960, except for
R1080* and R2488*, which lasted until 1965. According to CD, some of these
associations also operated collective discrimination arrangements before
1956. Despite the number of agreements, the combined share of the affected
products was only about 40–50% of total industry sales revenue—or less, if
at least some of the agreements did not cover the export market, which was
important in this industry.

Hand Tools and Implements (MLH 391)

The register contains price agreements by the Association of UK Plier Man-


ufacturers (R121), the File Trade Association (R485), the Light Edge Tool and
Allied Trades Association (R486*), the Scythe, Sickle and Hook Manu-
facturers Association (R487), the Precision File Association (R905), the Edge
Tool Manufacturers Association (R1559), the Flexible Back Bandsaw Manu-
facturers Association (R413), the Band Saw Association (R888), the Tubular
Frame Saw Association (R904), the Short Saw and Crosscut Saw Association
382 Appendix A

(R948), the Circular and Long Saw Association (R977), and the British Hack-
saw Makers Association (R1497, R1670). Some of these agreements were also
mentioned under MLH 390. According to CD and BT, some products of this
industry were subject to collective discrimination arrangements before 1956.
All the price agreements were abandoned in 1959, leading to a change of
competition regime in the hand tool industry.

Cutlery, Spoons, Forks, and Plated Tableware, etc. (MLH 392)

Razors and blades. No evidence of restrictive agreements.

Knives, tableware, etc. The Machine Knife Manufacturers Association set


prices for guillotine paper knife blanks until 1959 (R947). According to a case
study of the industry, written by H. Townsend and published in Burn (1958),
there were also pricing arrangements for certain other semifinished goods,
but not for finished goods. The affected products accounted for about 10–
20% of sales revenue in this product category.

Bolts, Nuts, Screws, Rivets, etc. (MLH 393)

A large number of agreements were registered, covering nearly all the


products of the industry. In particular, the register contains price agreements
by the Heated Bolt Association (R350*), the Bright Bolt and Nut Manufac-
turers Association (R408*), the Black Bolt and Nut Manufacturers Associa-
tion (R351*), the Aircraft Bolt and Nut Manufacturers Association (R348*),
the Stainless Steel Bar Products Association (R87), the Rubber Thread Screw
Association (R289), and various other associations of producers of rivets
(R280, R290, R607, R382), washers (R291), cotter pins (R306), and tacks, nails,
and related products (R375, R382, R492, R1066, R1069, R1070, R1071). There
were also agreements between some of these associations to observe each
other’s prices, and agreements between manufacturers and distributors,
relating mostly to trade and quantity discounts. According to CD, some
products of MLH 393 were subject to collective discrimination arrangements
before 1956. All price agreements ended at various dates between 1959 and
1968, except R351*, which was contested by the parties and was allowed by
the Court to continue (LR 2 RP 50, LR 2 RP 433, LR 3 P 43, LR 6 RP 1).
To summarize, the industry was collusive in the 1950s, and the combined
share of the products with terminated agreements in total industry sales
revenue was probably higher than 50%. On the other hand, black bolts and
nuts accounted for more than 20% of industry sales revenue, so the state of
competition in the industry as a whole was somewhat uncertain in the 1960s
and the 1970s. Not only did the price agreement in black bolts and nuts con-
tinue until at least the mid-1970s, but the firms involved were also members
of several other associations with formally abandoned agreements. The con-
tinuation of an explicit price agreement in one market may have reduced
Survey of Collusive Agreements in British Industries 383

competitive pressure in other, closely related, markets. For these reasons,


the effect of the 1956 Act on MLH 393 during the period examined in this
book should probably be seen as ambiguous.

Wire and Wire Manufactures (MLH 394)

Wire and wire manufactures of iron and steel. A large number of price
agreements by various associations and groups of wire manufacturers were
registered, covering all types of iron and steel wire and wire manufactures.
Most of these covered specific types of wire or wire manufactures, so it
would be rather tedious to give details on each here.5 According to CD,
several also provided for collective discrimination. Three agreements were
contested in the Court and struck down (see LR 5 RP 146): R867*, by the
Locked Coil Ropemakers Association; R869*, by the Wire Rope Manufac-
turers Association; and R993*, by the Mining Rope Association. Most agree-
ments in this part of MLH 394 were abandoned between 1959 and 1962,
although some, including R867*, R869*, and R993*, were canceled in 1964–
1965. In some cases, information agreements were in operation in the 1960s.
The slow emergence of competition in wire ropes is described in Swann et al.
(1973) and in MC, Wire and Fibre Ropes (London: H.M.S.O., 1973). On the
whole, the 1956 Act caused a change of competition regime in this part of the
wire industry in the long run.

Wire and wire manufactures of copper, brass, and other copper alloys.
The main source of information for this part of MLH 394 is MRPC, Report on
the Supply and Export of Certain Semi-manufactures of Copper and Copper-based
Alloys (London: H.M.S.O., 1955). This report described the collusive activ-
ities of several associations of producers of copper and copper alloy semi-
manufactures, including wire and wire manufactures. See the discussion
under MLH 322. Most of the associations investigated by the MRPC did not
register their agreements. However, the agreements of the Cadmium-Copper
and Bronze Association (R1554) and the High Conductivity Copper Associ-
ation (R508*) were registered and ended in 1959; both covered certain types
of wire. The register also contains two agreements on electric cable wire
(R630 and R634), both abandoned in 1962. To summarize, the 1956 Act
caused a change of competition regime in this part of the wire industry.

5. They are the following: R66, R67, R88, R148, R368, R369, R616, R623, R624, R627,
R628, R629, R631, R632, R633, R635, R636, R637, R638, R659* (by the Mild Steel Wire
Manufacturers Association), R866, R867*, R869*, R945, R993*, R995, R996, R998,
R1004, R1005* (by the Patented Steel Wire Association), R1031* (by a group of manu-
facturers of wire nails), R1067, R1068* (by the Reinforcement Conference), R1072* (by
the British Wire Netting Manufacturers Association), R1190, R1198, R1281, R1303,
R1314, R1377, R1389, R1436, R1845, R1846, R2313, R2981, R3072 (the last of these was
an effective information agreement of the Federation of Reinforcement Fabric Makers).
Some of these may also cover wire and wire products of copper, brass, and other
copper alloys.
384 Appendix A

Cans and Metal Boxes (MLH 395)

A price agreement on milk cans by members of the Can Manufacturers


Association was abandoned in 1958 (R815). This is a minor product of MLH
395, however. PEP mentions that the Tinbox Manufacturers Association
operated an agreement for buying prices of tin scrap, but it is not clear why
this should significantly affect competition between manufacturers in the
product market. Finally, an agreement between the largest producer of metal
boxes, Metal Box, and the largest producer of crown corks, Crown Cork,
specifying that each firm would not enter the industry of the other, ended in
1963 (R3110). Since this was not an agreement between manufacturers of
metal boxes, however, the industry can probably be regarded as competitive
for our present purposes.

Jewellery and Precious Metals (MLH 396)

There is no evidence of any restrictive agreements in this industry.

Metal Furniture (MLH 399.1)

Until 1960, the Metal Bedstead Association fixed minimum prices for metal
bedsteads, wire and spring mattresses, and other associated products (R1226*).
Moreover, the Bedstead Fittings Association set the prices of bedstead fit-
tings until 1959 (R1251). These products accounted for less than 10% of the
metal furniture industry throughout the 1960s and the 1970s. Office furniture
was subject to some agreements that will be discussed under MLH 472; none
of them was particularly restrictive. Hence the metal furniture industry can
be seen as competitive.

Drop Forgings, etc. (MLH 399.5)

The National Association of Drop Forgers and Stampers made recommen-


dations for increases in the price of drop forgings until 1965 (R114*, R115).
Certain products of MLH 399.5 were also covered by the pricing agreement
of the National Forgemasters Association (R1122*). This ended in 1962, and it
is not clear whether R2801*, R2802*, and R2803*, which essentially replaced
R1122* until 1965, covered products of MLH 399.5. In any case, there was a
change of competition regime in the long run in this industry.

Metal Hollow-ware (MLH 399.6, 399.7)

Domestic hollow-ware. The register contains price agreements by the


British Aluminium Hollow-ware Manufacturers Association (R204), the
Wrought Hollow-ware Trade Employers Association (R262), the Galvanised
Survey of Collusive Agreements in British Industries 385

Hollow-ware Association (R541), and the Dustbin Association (R639, R2321).


All the agreements were abandoned in 1959–1960, but R541 was replaced by
an information agreement, which ended in 1965. This industry experienced a
change of competition regime.

Industrial hollow-ware. A price-fixing agreement by the Association of


Steel Drum Manufacturers, covering steel drums and kegs, was canceled
in 1960 (R135*); it was replaced by an effective information agreement, which
lasted until 1965. See Swann et al. (1973) for details on the evolution of
competition in this industry. A price agreement between members of the
Galvanised Tank Association (R354*), covering galvanized tanks, cisterns,
and cylinders, was abandoned in 1959. According to CD, the association also
practiced collective exclusive dealing before 1956. After 1959, the industry
operated an information agreement, which was referred to the Court in 1965
and was found to have amounted to the same effect as the former explicit
pricing arrangement (LR 5 RP 315). The information agreement ended in
1965. All in all, the 1956 Act caused a change of competition regime in this
industry.

Miscellaneous Metal Goods (MLH 399.2, 3, 4, 8, 9, 10, 11 and 12)

Metal windows and door frames. The British Metal Window Association
fixed the prices of metal windows, door frames, and associated products
(R835, R2618*). In its original form, the agreement also provided for some
allocation of work. See Swann et al. (1973) and MRPC, Report on the Supply of
Standard Metal Windows and Doors (London: H.M.S.O., 1957). The agreement
weakened somewhat in the late 1950s and early 1960s, because the market-
sharing provisions were dropped and the leading producer of metal win-
dows resigned from the association. It was referred to the Court in 1962,
contested by the parties, and allowed to continue (LR 3 RP 198). However, it
was voluntarily abandoned shortly afterward. The register also contains a
price agreement by the Patent Glazing Conference (R836), which ended in
1959. Hence there was a change in competition regime in this industry.

Safes, locks, latches, and keys. The register contains price agreements
by the British Lock and Latch Manufacturers Association (R119) and the
Cylinder Lock Manufacturers Association (R187). They covered keys, locks,
and latches, and were abandoned in 1958–1959. Also relevant is the price
agreement of the National Brassfoundry Association, which covered a wide
range of brass goods, including locks and latches, and was canceled in 1960
(R349). The 1956 Act caused a change of competition regime in this industry.

Springs. A price agreement by the Spring and Interior Springing Associa-


tion, covering mattress and upholstery springs, was abandoned in 1960
(R1132*). Two other price agreements covering railway springs, by the Heavy
386 Appendix A

Coil Spring Association (R994*, R1321) and the Laminated Railway Spring
Association (R1002*, R1003), respectively, were abandoned in 1965. The
combined share of these products in total industry revenue was probably
higher than 50% (at least on the basis the 1968 S.I.C. definition of the industry,
which excludes springs for motor vehicles). Hence there was a change of
competition regime in this industry.

Needles, pins, and other metal small ware. The register contains price
agreements by the Pin and Allied Trades Association (R390), the Metal
Skewer Association (R202), the Machinery Belt Fastener Manufacturers As-
sociation (R2948) and two groups of producers of pins and similar articles
(R2292, R2641). All the agreements were abandoned between 1958 and 1962,
although R390 was replaced for some time by an information agreement. BT
also mentions slide fasteners (zippers) as an allegedly collusive product. The
combined share of the affected products was about 30% of total industry
sales revenue. Several products of the industry were also listed in CD.

Domestic gas appliances. The register contains an agreement between the


Society of British Gas Industries (representing manufacturers of most types
of domestic gas appliances) and the Gas Council (representing the Area
Boards, the principal buyers of appliances) (R823). Under the agreement the
Boards were to sell domestic gas appliances at the list prices published by
the manufacturers, and the manufacturers were to grant to the Boards speci-
fied discounts and not to give wholesalers and retailers discounts higher than
those given to the Boards. The agreement ended in 1960. However, its impli-
cations for competition in the particular circumstances of an industry with a
dominant buyer are not clear. In fact, the agreement seems designed to pro-
tect the interests of the Area Boards as much as those of the manufacturers.
Moreover, the long-run effect of the abolition of the agreement is uncertain.
An official investigation into the gas appliance industry carried out in the
late 1970s revealed several restrictive agreements, relating mainly to dis-
counts, the exchange of price information, and conditions of sale, which had
been in operation for various dates since the early 1970s (R4951, R4956–
R4967). Further information is contained in MMC, Domestic Gas Appliances
(London: H.M.S.O., 1980). Those arrangements that were still in force were
abandoned in the late 1970s. In conclusion, the state of competition in this
industry was probably ambiguous in the 1950s and/or the 1960s and 1970s.

Metallic closures. The register contains an information agreement among


three important firms in the crown cork industry, which resulted in occa-
sional informal understandings about price changes (R3131). It is not clear
how long the agreement had been in force and whether it was a replacement
for a formal price-fixing scheme. The information agreement was discon-
tinued in 1968. Crown corks accounted for about 25% of the total sales of
metallic closures in 1963.
Survey of Collusive Agreements in British Industries 387

Metal finishing. The register includes price-fixing agreements by the As-


sociation of London Galvanisers (R205), the Vitreous Enamellers Association
(R792), the Association of Galvanised Steel Gutter and Pipe Manufacturers
(R831), the Guild of Metal Perforators (R609), the Association of Scottish
Galvanisers (R604), and the Sheffield District Annealers and Heat Treaters
Association (R1310). All of these were canceled in 1959–1960. They probably
covered about a third of total sales of the industry. PEP also reports (on the
basis of evidence given by a trade association of galvanizers) some local
pricing agreements on zinc galvanizing.

Other metal goods. A price agreement by the National Brassfoundry As-


sociation, covering a wide range of brass goods and including ‘‘engineers’
and mechanicians’ goods,’’ was abandoned in 1960 (R349). A price-fixing
agreement by the Stockless Anchor Association ended in 1959 (R1636); how-
ever, stockless anchors represented less than 10% of total sales of the product
category ‘‘anchors and chains.’’ The register also contains price agreements
by the Steel Hinge Makers Association (R1028*, R2259), the Scottish Associ-
ation of Manufacturing Coppersmiths (R901), the Umbrella Components
Association (R1006), the Hook and Band Manufacturers Association (R2506),
the Ships Ordinary Sidelights Association (R480), and groups of producers
of metal clips and brackets (R1496), corrugated joint fasteners (R1353), and
hammerlock struts (R2664). All these agreements were canceled between
1959 and 1961. CD also reports that copper cylinders and boilers, radiator
blocks, flushing cisterns and copper balls, curtain rods, and other hardware
products were subject to collective discrimination arrangements before 1956.

Production of Man-Made Fibres (MLH 411)

No agreements were registered in this industry. The evolution of competi-


tion during the 1950s is, however, described in MC, Man-made Cellulosic
Fibres (London: H.M.S.O., 1968). In the viscose section (about 40% of total
industry sales revenue in 1958), discussions on prices between the dominant
producer, Courtaulds, and a number of smaller firms were discontinued in
1956. In the acetate section (about 10% of total industry sales revenue in
1958), price discussions were less effective than in viscose. Moreover, the two
significant producers, Courtaulds and British Celanese, merged in 1957, so
the 1956 Act would have little effect after 1958 anyway. Finally, in nylon, the
only producer was British Nylon Spinners, a firm jointly owned by Cour-
taulds and the chemical giant I.C.I. The monopoly position of B.N.S. was
protected by an agreement whereby the parent firms would produce nylon
only through B.N.S. and, in return, B.N.S. would produce only nylon. See
also D. C. Coleman, Courtaulds: An Economic and Social History, vol. 3 (Ox-
ford: Clarendon Press, 1980). The implications of the agreement between
Courtaulds and I.C.I. are not clear, however. Since there was only one firm
producing nylon, there could be no collusion specifically in nylon. Even if
one looks at the market for man-made fibres as a whole, the fact that Cour-
388 Appendix A

taulds had a 50% share in B.N.S. complicates any assessment of the con-
sequences of the Courtaulds-I.C.I. agreement for competition and for the
structure of the industry. Courtaulds was also a party to a number of inter-
national collusive agreements since 1950; these are not, however, relevant for
our present purposes.
In conclusion, only the viscose section of the industry can be regarded as
unambiguously collusive. The share of viscose sales in total industry sales
was less than 50% in the late 1950s, and declined in the 1960s.

Spinning and Doubling on the Cotton and Flax Systems (MLH 412)

Single yarn of cotton, glass fibres and man-made fibres. An agreement


by the Yarn Spinners Association contained minimum prices for cotton yarn
(R79*). The agreement was abandoned in 1959, after being unsuccessfully
contested in the Restrictive Practices Court (LR 1 RP 118). A price agreement
by the Rayon Staple Spinners and Doublers Association, covering yarn of
man-made staple fibers, ended in late 1958 (R1137). There was therefore a
change of competition regime in this industry.

Doubled yarn of cotton, glass fibres, and man-made fibres. The Cotton
Yarn Doublers Association set minimum prices for doubled cotton yarn until
1959 (R269*). And the Rayon Staple Spinners and Doublers Association
regulated the price of doubled yarn of man-made staple fibers until 1958
(R1137). However, the price of doubled yarn of man-made continuous fila-
ment apparently was not regulated. The combined share of the products
covered by the two agreements in total industry sales revenue was about
50% in 1958, but only 35% in 1963 and 25% in 1968.

Finished thread. The Linen Sewing Thread Manufacturers Association


specified quotas for sales of linen thread to government departments and
nationalized industries, and also set minimum prices for certain qualities of
linen thread (R422). According to the association, the pricing restrictions
covered a relatively small fraction of total sales of linen thread. The agree-
ment ended in 1960, and was replaced by an information agreement, which
continued to operate until the early 1970s. Total sales of linen thread
accounted for slightly more than 10% of total sales of finished thread in the
late 1950s and early 1960s, but for less than 10% in the late 1960s and the
1970s. The finished thread industry should probably be seen as competitive.

Flax. An agreement by a group of Scottish flax spinners fixed the prices of


flax line and tow yarns until 1958 (R251). Scotland accounted for less than
40% of the total UK production of flax yarns in the 1950s.

Other products and work done. Until 1959, the Precision Winding Asso-
ciation fixed the charges for electrical-type winding of yarn (R916). A group
Survey of Collusive Agreements in British Industries 389

of silk spinners allocated supplies of silk waste through a common buying


organization until 1960 (R568). Finally, an agreement by the Soft Hemp and
Tow Spinners Association, comprising prices for soft hemp and hemp tow,
was abandoned in 1960 (R1221). PEP also mentions the Silk Throwsters As-
sociation as a potentially collusive body.

Weaving of Cotton, Linen, and Man-Made Fibres (MLH 413)

Woven cotton cloth in the loom state. The register contains three agree-
ments, all of which related to minor products of the industry. An agreement
by the Ventile Fabrics Association of Great Britain, whose membership con-
sisted of spinners, doublers, weavers, and finishers, required members to
deal with each other only (R403); this was discontinued in 1960. A similar
agreement by the Cotton Velvet Council, whose membership consisted of
manufacturers and finishers, was canceled in 1957 (R92). Finally, the Win-
dow Holland Association, which consisted of manufacturers and finishers,
set prices and also operated an exclusive dealing arrangement between the
members until 1959 (R78, R1253). All these were secondary products of the
industry, however, so the influence of these agreements on the industry as
a whole would have been small. According to evidence given by a cotton
weaver and reported in PEP, attempts to reach an agreement between firms
covering the main products of the industry had consistently failed.

Woven cloth of man-made fibres in the loom state. An agreement by the


Rayon Weaving Association provided for minimum prices and recom-
mended price increases for woven cloth of rayon until 1959 (R1138). The
affected products covered more than 50% of industry sales revenue. In
other words, the 1956 Act caused a change of competition regime in the
industry.

Woven cloth of flax or hemp in the loom state. The Association of Flax
Canvas Weavers set the prices of certain categories of flax and/or hemp
canvas (R680). In three separate agreements, certain groups of weavers
agreed upon prices of flax canvas for sale to government departments (R679,
R1487, R1851). These agreements ended in 1959–1960. The products affected
accounted for more than 20% but less than 50% of total industry sales.

Other products. The Cotton Canvas Manufacturers Association specified


minimum prices for rayon tire cord and for cotton canvas fabrics (R31). An
agreement by the Canvas Hose Manufacturers Association specified prices
for canvas hose (R560). The Book Cloth Association set prices for book cloth
(R103). All these agreements were abandoned in 1959–1960. Moreover, until
1958, the Silk Trade Employers Association fixed the prices of certain types
of silk fabrics (R244). The Association of Cleaning Waste Manufacturers
390 Appendix A

operated a market-sharing agreement until 1959 (R2113). After 1959, this was
replaced for some time by a quantity information agreement. Finally, an
agreement by the British Fabric Federation, relating only to conditions of sale
for woven and knitted fabric for wearing apparel, was abandoned in 1967
(R2993*).

Woollen and Worsted (MLH 414)

Wool and other animal hair sorted, blended, scoured, or carbonized. A


price-fixing agreement by the Wool Carbonisers and Scourers Federation
continued until 1969 (R3148*). The parties to this agreement were firms that
specialized in the scouring and carbonizing of wool. As pointed out by Hart
et al. (1973), this type of specialized activity was common in the past, but
over time vertical integration in the woolen and worsted industry resulted
in a substantial shift of market share away from independent scourers and
toward integrated topmaker-combers. Specialist firms accounted for much
less than 50% of total sales revenue in the industry in the late 1960s (i.e., at
the time when the federation abandoned its agreement).

Combing of wool tops. There is no evidence of any significant restrictive


agreements among integrated topmaker-combers, who were responsible for
most combing of wool in the 1960s. Thus the 1956 Act had no effect on this
section of the industry. On the other hand, the register contains two price-
fixing agreements by associations of commission combers, the Commission
Woolcombers Association (R1177) and the Commission Recombers Associa-
tion (R1176). Both agreements were abandoned in 1959, leading to a change
of competition regime in this (much smaller) section of the wool combing
industry.

Yarn of wool, hair, and man-made fibres spun on the worsted system.
An agreement by the Branded Knitting Wool Association, abandoned in
1960, comprised discounts for hand-knitting wool (R1926). A registered
agreement by the Worsted Spinners Federation, relating only to terms of
contract and settlement discounts, was allowed to continue without a Court
hearing (R1569). According to PEP, however, the federation recommended
minimum prices in the 1950s.

Yarn of wool, hair, and man-made fibres spun on the woollen system.
An agreement by the Scottish Woollen Spinners Association, containing
some minor pricing restrictions, was canceled in 1960 (R218). Neither this
nor the agreement of the Branded Knitting Wool Association described
above (R1926) must have had any significant effect on the industry as a
whole: in both cases the restrictions were minor and the fraction of total in-
dustry sales covered was small. The same can be said about the agreements
Survey of Collusive Agreements in British Industries 391

on yarn for the Harris Tweed industry (see below). Hence the woolen yarn
industry was essentially competitive.

Woven worsted fabrics. No evidence of any significant restrictive agree-


ments. According to PEP, the Woollen and Worsted Manufacturers Associa-
tion only specified discounts.

Woven woollen fabrics. The Hebridean Spinners Advisory Committee and


the Harris Tweed Industry Consultative Committee set the prices of yarn
and Harris tweed until 1964 (R1223*, R2760*). This is a minor product of the
industry, however, so the influence of these agreements on the industry as a
whole would have been small. The same is probably true for an agreement
between a small group of manufacturers of woolen cloth and serge cloth,
which specified prices to be charged to government departments (R737). This
industry was therefore largely competitive.

Blankets. An agreement by the Blanket Manufacturers Association specified


a minimum price for a basic quality of woolen blankets (R801*). The agree-
ment was abandoned in 1959, after being unsuccessfully contested in the
Court by the parties (LR 1 RP 208). The effect of the agreement on competi-
tion in the industry is not clear, however, since the price restriction covered
only a certain type of blanket and—as pointed out in the Court report—no
sales were actually being made at the minimum prices. Hence the effect of
the 1956 Act on the blanket industry was ambiguous.

Other products and work done. The British Paper Machine Felt Associa-
tion set the price of felt for industrial machines (R1534). This agreement was
discontinued in 1959, and was replaced by an information agreement. The
Lancashire Mechanical Cloth Manufacturers Association fixed the price of
cloth for industrial machines until 1960 (R917). The Moquette Manufacturers
Association fixed the prices of moquettes for the railways until 1959 (R590).
Finally, the Bradford and District Waste Pullers Association specified charges
for wool shoddy until 1966 (R3032).

Jute (MLH 415)

The register contains a series of price agreements covering all the main sec-
tions of the jute industry: spinners, weavers, merchants, and importers. In
particular the agreements covered jute yarn (R1340*), jute cloth (R1339*),
hessian (burlap) piece goods (R1337*, R1338*), and jute carpets (R1345*). The
industry was also listed in CD. The first four of these agreements were un-
successfully contested in the Court. All were abandoned in 1963, but they
were replaced by information agreements, which were effective for some
time (see Hart et al. 1973, Swann et al. 1974). There was, however, a change
of competition regime in the long run in the jute industry.
392 Appendix A

Rope, Twine, and Net (MLH 416)

The main products of this industry were subject to tight collusion in


the 1950s. According to MRPC, Report on the Supply of Hard Fibre Cordage
(London: H.M.S.O., 1956), this involved common prices, market sharing,
aggregated rebates, collective exclusive dealing, collective resale price main-
tenance, and international agreements limiting imports. The industry was
also listed in CD and BT. Only watered-down versions of the original collu-
sive arrangements in the industry were registered. In particular, the register
contains price-fixing agreements by the Hard Fibre Rope Manufacturers As-
sociation (R617*), the Trawl Twine Manufacturers Association (R618*), the
Hard Fibre Cord and Twine Manufacturers Association (R619*), the Syn-
thetic Cordage Manufacturers Association (R622), the Agricultural Twine
Manufacturers Association (R626*), the Plaited Cordage Association (R1231),
the Twine Manufacturers Association (R1563, R1797), the Scottish Net Man-
ufacturers Association (R1356, R2736), and a group of producers of com-
bined wire and fiber rope (R865, R1302). All these agreements were
abandoned in 1959–1960. According to MC, Wire and Fibre Ropes (London:
H.M.S.O., 1973), competition in the early 1970s was not fully effective in all
markets within the industry. Competition further intensified as a result of
action taken after the publication of the MC report. All in all, there was a
change of competition regime in MLH 416.

Hosiery and Other Knitted Goods (MLH 417)

Knitted, netted, or crocheted fabrics. The Silk and Rayon Trade Protec-
tion Society, whose membership consisted of weavers, finishers, and mer-
chant converters of various types of fabric, recommended minimum prices
for warp-knitted fabrics and also for certain finishing processes until 1961
(R1637). The affected products accounted for more than 10% but less than
50% of the knitted fabric industry. A price agreement by the Stockinette
Manufacturers Association was abandoned in 1959 (R1448); it covered only a
secondary product of the industry, however. Finally, the agreement of the
British Fabric Federation described under MLH 413 (R2993*) covered prod-
ucts of MLH 417 as well, although it was not particularly restrictive.

Hosiery. No evidence of any restrictive agreements. An agreement by the


National Hosiery Manufacturers Federation contained only conditions of
sale and was allowed to continue without a Court hearing (R1200).

Underwear, shirts, and nightwear, knitted, netted, or crocheted. No evi-


dence of any restrictive agreements.

Other clothing, knitted, netted, or crocheted. No evidence of any restrictive


agreements.
Survey of Collusive Agreements in British Industries 393

Lace (MLH 418)

The Madras Manufacturers Association specified prices for madras goods


until 1959 (R430). This is not, however, a major product of the lace industry.
More important, the manufacturers of leaver’s lace agreed to sell all their
output through a common selling organization, which set minimum prices
(R3163*). These restrictions lasted until 1970. Leaver’s lace accounted for
about 10–25% of total sales of the lace industry during the period examined
here.

Carpets (MLH 419)

Woven carpets, carpeting, and carpet floor rugs. A series of agreements


by the Federation of British Carpet Manufacturers provided for minimum
prices for certain types of woven carpets, notification of price changes, trade
and quantity discounts, conditions of sale, and resale price maintenance
(R1316*, R1317*, R1318*, R1319*, R1320*). The industry was also listed in
CD. The most important of these agreements were contested in the Court
and were struck down (LR 1 RP 472). According to Swann et al. (1973),
competition intensified in the 1960s despite parallel pricing and the exchange
of price information. The 1956 Act caused a change of competition regime in
this section of the carpet industry.

Tufted carpets, carpeting, and carpet floor rugs. No evidence of any re-
strictive agreements.

Other carpets, carpeting, and carpet floor rugs, and pile fabric rugs.
Some products that fall into this category were covered by the agreements of
the Federation of British Carpet Manufacturers described above. Moreover,
until 1961 a series of registered agreements by the Association of Manu-
facturers of Mohair and Pile Floor Rugs and Mats provided for specified
trade discounts, the maintenance of manufacturers individual prices, agreed
prices for comparable qualities, and no changes in individual prices without
prior discussion with other members (R557, R1831–R1835).

Narrow Fabrics (MLH 421)

Elastic and elastomeric goods. The register contains a price agreement


by the British Elastic Braid Manufacturers Association, which was canceled
in 1959 (R143). The affected products accounted for 10–20% of total sales
revenue in elastic and elastomeric goods.

Woven machinery belting and other narrow fabrics. An agreement by


the Tape Manufacturers Association specified minimum prices for standard
394 Appendix A

cotton and electrical tapes until 1962 (R562*). These products accounted
for 10–20% of total sales revenue in this part of MLH 421. In addition, BT
reports allegations of collusion between manufacturers of conveyor belting
and collective exclusive dealing in the brace webbing industry.

Made-Up Household Textiles and Handkerchiefs (MLH 422.1)

The register contains two agreements, neither of which seems particularly


important for the industry as a whole. The first was a market-sharing agree-
ment between five suppliers of rented linen goods; it was abandoned in 1967
(R3104*). The second was a pricing agreement by the Traced Art Needlework
Manufacturers Association, which ended in 1961 (R188). The evidence re-
ported in PEP confirms that MLH 422.1 was a competitive industry.

Canvas Goods and Sacks and Other Made-Up Textiles (MLH 422.2)

Canvas sacks and bags. Jute sacks and bags were covered by the agree-
ments between producers of hessian (burlap) piece goods described under
MLH 415 (R1337*, R1338*). In addition, the Jute Sack and Bag Manufacturers
Association recommended charges for the sewing, stamping, and bundling
of jute sacks and bags (R1336*). Merchants and importers were also parties
to the pricing arrangements (R1334*, R1335*). All these agreements were
contested in the Court, struck down, and abandoned in 1963. Although they
were replaced for some time by information agreements, the industry expe-
rienced a change of competition regime in the long run as a result of the 1956
Act.

Made-up goods of sailcloth, canvas, and the like. An agreement by a group


of producers of canvas freight car covers, relating to prices and the allocation
of work, ended in 1961 (R2680). The affected products probably accounted
for 10–20% of total sales revenue in this part of MLH 422.2.

Textile Finishing (MLH 423)

The register contains about thirty restrictive agreements by various associa-


tions and groups of textile finishers, covering all types of textiles and finish-
ing processes. I will not provide details on each of these agreements here.6
All of them were mainly about prices, and some had originally comprised
additional restrictions, such as market sharing—see MRPC, Report on the
Process of Calico Printing (London: H.M.S.O., 1954). All were abandoned be-

6. Their numbers in the register are as follows: R75, R170, R200, R277, R284, R367,
R393* (by the Dyers and Finishers Association), R402, R682, R830, R838, R942, R979,
R1160, R1245, R1315, R1324, R1531, R1533, R1535, R1567, R1568, R1582, R1606, R1612,
R1637, R1705, R1800, R1801, R2212, R2600.
Survey of Collusive Agreements in British Industries 395

tween 1958 and 1962, although several were initially replaced by information
agreements, some of which lasted until 1970. There was therefore a change of
competition regime in MLH 423 as a result of the 1956 Act.

Asbestos Manufactures (MLH 429.1)

The evolution of competition in this industry is described in MC, Asbestos


and Certain Asbestos Products (London: H.M.S.O., 1973). Prior to 1956, the
industry operated price-fixing agreements in most major products, such as
asbestos textiles, jointings and related products, brake and clutch linings,
and asbestos insulation products. (Brake linings were also listed in BT as
allegedly collusive.) For some products there were market-sharing provi-
sions as well. These agreements were not registered. Informal discussions on
prices and discounts continued until the late 1960s, although they were not
necessarily as effective as the agreements of the 1950s. Some of these arrange-
ments were discovered in the late 1960s and were placed on the register
(R3774, R3708). By the early 1970s, all restrictive arrangements apparently
had ended. Although the emergence of competition in this industry was
delayed for over a decade after the introduction of the 1956 Act, by the mid-
1970s the industry had probably experienced a change in competition regime.

Miscellaneous Textiles (MLH 429.2)

A price agreement by the British Mat and Matting Manufacturers Associa-


tion, covering coir mats and mattings, was abandoned in 1959 (R253). An
agreement by the Kapok Processors Association specified minimum prices
for pure kapok and kapok mixtures until 1957 (R51). BT also reports allega-
tions of collective exclusive dealing in processed kapok. The products covered
by the two agreements accounted for somewhat less than 20% of total sales
revenue of MLH 429.2 as a whole.

Leather (Tanning and Dressing) and Fellmongery (MLH 431)

Leather, undressed. The register contains an agreement by the Pigskin


Tanners Federation, whose membership consisted of seven Scottish tanners
(R2682*). The agreement related to buying prices and buying quotas for
pigskins and hogskins, and was abandoned in 1963. This was a regional
agreement on a specific product. The 1956 Act had no significant effect on the
leather tanning industry as a whole.

Leather, dressed. No evidence of important restrictive agreements, other


than allegations of collective discrimination arrangements reported in BT.

Fellmongery. The register contains a number of agreements between two


regional associations of sheep hide and skin sellers and two regional associ-
396 Appendix A

ations of fellmongers (buyers of sheep hide and skin) (R2690, R2691, R2695).
Under the agreements, each sellers’ association should not sell outside its
territory and each buyers’ association should buy all the sheepskins offered
to it at specified prices. The agreements were abandoned between 1959 and
1961, but they are unlikely to have had any significant impact on competition
between fellmongers. The fellmongery industry was essentially competitive.

Manufactures of leather not elsewhere specified. The Buffalo Picker Manu-


facturers Association set prices for buffalo hide pickers until 1959 (R431).

Leather Goods (MLH 432)

There is no evidence of restrictive agreements in this industry.

Fur (MLH 433)

There is no evidence of restrictive agreements in this industry.

Weatherproof Outerwear (MLH 441)

There is no evidence of any significant restrictive agreements in this industry.


An agreement by the Clothing Manufacturers Federation of Great Britain,
which covered several clothing industries (including MLH 441), is discussed
under MLH 443.

Men’s and Boys’ Tailored Outerwear (MLH 442)

There is no evidence of any significant restrictive agreements in this industry.


An agreement by the Clothing Manufacturers Federation of Great Britain,
which covered several clothing industries (including MLH 442), is discussed
under MLH 443.

Women’s and Girls’ Tailored Outerwear (MLH 443)

The register contains agreements by the Clothing Manufacturers Federation


of Great Britain (R546*), the British Mantle Association (R2464*), and the
Apparel and Fashion Industry’s Association (R1341*). The three agreements
were similar. All three contained a restriction that fixed the seasonal dates at
which women’s outerwear could be sold at reduced prices. In addition,
R546* and R1341* required that members not grant any discounts to buyers
other than settlement discounts, and contained a few more minor conditions
of sale. R2464* was canceled in 1964, while R546* and R1341* were aban-
doned in 1968. It is unlikely, however, that these agreements had any sig-
nificant effect on competition. This view is supported by evidence provided
Survey of Collusive Agreements in British Industries 397

in the 1950s by firms in the clothing industry and reported in PEP: according
to one firm, the restrictions on discounts and conditions of sale of the Clothing
Manufacturers Federation were not observed; according to another, the
restrictions on discounts and conditions of sale of the Apparel and Fashion
Industry’s Association were observed, but competition was nevertheless
very intense.

Overalls and Men’s Shirts, Underwear, etc. (MLH 444)

There is no evidence of any significant restrictive agreements in this industry.

Dresses, Lingerie, Infants’ Wear, etc. (MLH 445)

There is no evidence of any significant collusive agreements in this industry.


The agreement of the Apparel and Fashion Industry’s Association, described
under MLH 443, covered this industry as well, but, as pointed out above, it
did not contain any significant restrictions on competition. Another agree-
ment, by the Light Clothing and Allied Trades Association, comprised set-
tlement terms, delivery charges, and sample discounts (R652, R625). It ended
in 1964, and was also of little significance for competition.

Hats, Caps, and Millinery (MLH 446)

The register contains agreements by the South of England Hat Manufac-


turers Federation (R275), the London Wholesale Millinery Manufacturers
Association (R804), the Millinery Guild (R1971*), and the Millinery Distrib-
utors Association (R312*). The four agreements covered women’s hats only.
R1971*, which contained restrictions about sales promotion and the seasonal
dates at which goods could be supplied to retailers at reduced prices, con-
tinued until 1968. The other three agreements were somewhat interrelated.
They provided for collective exclusive dealing, fixed distributors’ margins,
and conditions of sale and resale; see also CD. In addition, R275 comprised
recommendations for price increases, and R312* provided for minimum
prices of hats produced by wholesalers. All three agreements were aban-
doned in 1961. Given that the two manufacturers’ associations involved in
these agreements were regional and sales of ladies’ hats accounted for about
half of total sales of MLH 446 in the 1960s, one can conclude that the agree-
ments affected a fraction of total industry sales between 10% and 50%.

Corsets and Miscellaneous Dress Industries (MLH 449.1)

Corsetry. An agreement by the Corsetry Manufacturers Association about


settlement and sample discounts was allowed to continue without a Court
hearing (R542). This industry was clearly competitive.
398 Appendix A

Miscellaneous dress industries. An agreement by the Tie Manufacturers


Association, relating to trade discounts, as well as to prices for certain qual-
ities of ties, was abandoned in 1968 (R547*). The products covered by the
agreement accounted for 40–50% of total sales revenue in this product
group.

Gloves (MLH 449.2)

There is no evidence of restrictive agreements in this industry.

Footwear (MLH 450)

Leather footwear. No evidence of restrictive agreements.

Rubber footwear. The agreement of the Rubber Footwear Manufacturers


Association related to prices charged for rubber boots and certain other
standardized products, trade discounts, the maintenance of manufacturers’
resale prices, and collective exclusive dealing. See MRPC, Report on the Sup-
ply of Certain Rubber Footwear (London: H.M.S.O., 1956) for details. The in-
dustry was also listed in CD. The agreement presumably was abandoned in
1956, since it was not registered. Thus the industry experienced a change in
competition regime following the introduction of the 1956 Act.

Parts of shoes and work done. The Shoe Tip Association fixed the prices
of heel tips and toe plates until 1961 (R168). An agreement by a group of
manufacturers of cut leather soles provided, until 1959, for specified trade
discounts and aggregated rebates (R1825). Finally, BT mentions allegations
of collective exclusive dealing arrangements in boot and shoe repairing.

Bricks, Fireclay and Refractory Goods (MLH 461)

Refractory goods. The register contains price-fixing agreements by the


Welsh Silica Association (R69), the Sheffield Ganister and Compo. Associa-
tion (R224), the National Silica Brickmakers Association (R342), the Fireclay
Grate Back Association (R343), the Stourbridge Firebrick Association (R409,
R2650), the Woodville District Firebrick and Fireclay Association (R882), the
Yorkshire Firebrick Association (R833), the Scottish Firebrick Association
(R983*), the Magnesite and Chrome Brickmakers Association (R1625), and a
group of firebrick producers (R1089, R1090). In addition, a national organi-
zation, the National Firebrick Conference, made recommendations for price
increases in firebricks, fireclay, and fireclay refractories to cover changes in
costs (R341). All these agreements were abandoned between 1959 and 1961.
In some cases, they were replaced for some time by information agreements
Survey of Collusive Agreements in British Industries 399

(Swann et al. 1974). Bricks were also listed in CD. In summary, there was a
change of competition regime in this industry.

Building bricks. The register contains price-fixing agreements by several


regional associations of brick producers (R58, R151, R215, R489, R599*,
R1422). These registered agreements did not cover most areas. This is con-
sistent with the view expressed in a case study of the building materials in-
dustry, written by B. R. Williams and included in Burn (1958). On the other
hand, according to PEP, most areas were covered by restrictive agreements
in the mid-1950s. In addition, there were national pricing agreements for
specific types of bricks: by the Pressed Brick Makers Association (R1277),
which covered fletton bricks; by the Stock Brick Manufacturers Association
(R938); and by the Enamel Association (R642). According to PEP, R1277 and
R938 were effective primarily in the London area. All these agreements were
abandoned between 1958 and 1961. Given that fletton bricks accounted for
some 40% of total industry output in the 1950s and a significant part of the
rest of the industry was also cartelized, it is probably fair to say that the 1956
Act caused a change of competition regime in the building brick industry.
An official inquiry into the industry carried out in the 1970s found that
competition was effective at the time of the inquiry; see MMC, Building Bricks
(London: H.M.S.O., 1976).

Sanitary ware of fireclay, etc. The British Sanitary Fireclay Association


fixed prices and aggregated rebates for sanitary fireclay ware until 1958
(R1446). The Scottish Fireclay Pipe Association fixed prices for fireclay pipes,
drainage goods, and related products until 1960 (R1565, R1808). A pricing
agreement by the South Wales and Mon. Sanitary Pipe Association was
abandoned in 1960 (R1044, R1789). The industry was also listed in CD. Swann
et al. (1973) describe the emergence of competition following the termination
of collusive agreements. The 1956 Act caused a change of competition regime
in this industry.

Other products. Most other products of MLH 461 were subject to collusion
in the 1950s. In particular, the register contains price agreements by the Salt
Glazed Conduit Association (R885); the National Clayware Federation, cov-
ering a wide range of clayware goods (R877); the Bridgewater Roofing Tile
Manufacturers Association, covering roofing tiles of clay and fittings (R1646);
the South Eastern Brick and Tile Federation, covering clay bricks and roofing
tiles (R58); the Clay Block Association (R194); the Floor Quarry Association
(R573); and the National Horticultural Pottery Manufacturers Association
(R68). All these agreements ended in 1958–1959. Finally, according to the
case study of the building materials industry in Burn (1958), there was also a
price agreement for earthenware pipes. PEP also reports complaints that the
National Salt Glazed Pipe Manufacturers Association operated a price-fixing
agreement, and the salt glazed pipe industry was also listed in CD and BT.
400 Appendix A

Pottery (MLH 462)

Glazed earthenware tiles. A price-fixing agreement by the Glazed and


Floor Tile Manufacturers Association (R574, R2434, R2652*) was contested in
the Court in 1964 and was allowed to continue (LR 4 RP 239). The agree-
ment, which also provided for some degree of technical cooperation, covered
all types of ceramic wall, fireplace, and floor tiles, and was still in force in the
mid-1970s. According to CD and BT, the industry was also subject to collec-
tive discrimination arrangements before 1956. Swann et al. (1973) provide
details on the evolution of the industry from the 1950s to the early 1970s. The
glazed tile industry was collusive throughout the time period examined in
this book.

Sanitary earthenware. The British Sanitary Earthenware Manufacturers


Association fixed the prices of sanitary earthenware until 1958 (R254). Ac-
cording to CD and BT, it also operated collective discrimination arrange-
ments before 1956. Thus the 1956 Act caused a change of competition regime
in this industry. An official inquiry carried out in the 1970s found that de-
spite the occurrence of parallel pricing, competition was effective at the time
of the inquiry; see MMC, Ceramic Sanitaryware (London: H.M.S.O., 1978).

Other china and earthenware. A series of price agreements by members


of the Earthenware Association and between the association and a group of
outside firms, covering a wide range of items of earthenware, were aban-
doned in 1958 (R733, R791, R829). The British Teapot Manufacturers Associ-
ation fixed prices for teapots, jugs, and associated items of earthenware until
1962 (R894). The English China Manufacturers Association fixed prices for
bone china until 1958 (R1785). A case study of the pottery industry, written
by B. R. Williams and published in Burn (1958), mentions several other
associations of producers of china and earthenware with collusive agree-
ments in the 1950s, and notes that some of the agreements in this industry
were not as effective as the agreements in other sections of MLH 462. Nev-
ertheless, the china and pottery industry should be regarded as one with a
change of competition regime.

Electrical ware. An agreement by the British Electro-ceramic Manufac-


turers Association, covering die-made electroceramics for electrical insula-
tion, provided for prices and the reporting of inquiries (R956). It was canceled
in 1959. Hence there was a change of competition regime in this industry.

Glass (MLH 463)

Plate glass. A single producer, Pilkington Brothers, was responsible for the
entire production of plate glass. This firm participated in a number of restric-
Survey of Collusive Agreements in British Industries 401

tive agreements with distributors of plate glass, which were abandoned be-
fore or in 1956. After 1958, it was a party to an international market-sharing
arrangement. See CD and especially MC, A Report on the Supply of Flat Glass
(London: H.M.S.O., 1968). Unprocessed plate glass is the main product of
this part of MLH 463. Another important product is processed plate glass.
The register contains a price agreement by the Plate Glass Association, whose
membership consisted of processors and dealers in plate glass (R821*). This
agreement was abandoned in 1959, and was replaced by an information
agreement, which ended in 1965. The register also contains price-fixing agree-
ments by the Glass Benders Association (R837*), the Scottish Glass Mer-
chants and Glaziers Association (R1188) and the N. Ireland Plate Glass
Association (R827), all of which were cancelled in 1958–1960. R837* and
R1188 were replaced by information agreements, which lasted until the mid-
1960s. The share of processed plate glass in the total sales revenue of the
plate glass industry was 15–30% in the 1960s.

Safety glass. An agreement between two of the biggest producers of safety


glass, Pilkington and Triplex, relating to the demarcation of their respective
activities and exclusive cross-licensing, was abandoned in 1958 (R347). The
effect of this agreement on the industry as a whole is not clear.

Glass containers. The agreement of the Glass Bottle Association set mini-
mum prices for glass containers (R660*, R1681, R1682). It was contested in the
Court and struck down in 1961 (LR 2 RP 345). Swann et al. (1973) provide
details on this case. The 1956 Act caused a change of competition regime in
this industry.

Illuminating glassware. No agreements were registered in this industry.


However, the 1968 MC report on electric lamps provides some relevant in-
formation. The principal product of the industry in the 1950s was glass bulbs
for electric lamps. There were three major producers, one of which was a
joint subsidiary of the other two. Nothing is known about restrictive agree-
ments between these firms, but it is difficult to see this industry as competi-
tive. During the 1960s, the share of glass bulbs for electric lamps in total sales
of illuminating glassware declined substantially. Because of this change in
product mix and the ambiguity regarding the initial competitive conditions
in the industry, there is considerable uncertainty about the evolution of
competition in illuminating glassware.

Other glass products. Competition in the laboratory glassware industry


was regulated by the activities of two associations: the British Chemical
Ware Manufacturers Association, whose membership consisted of manufac-
turers of laboratory glassware, and the British Laboratory Ware Association,
which represented distributors that provided ‘‘technical service’’ and acted
as agents. An agreement between the two associations provided for collec-
402 Appendix A

tive exclusive dealing and specified trade discounts (R374). Moreover, an


agreement by the distributors’ association provided for common prices for
unbranded goods and the maintenance of individual prices of branded
goods (R654). The laboratory ware industry was also listed in CD and BT.
These agreements ended in 1959. The register also contains a price agree-
ment by the Stourbridge Glass Manufacturers Association, which was aban-
doned in 1959 (R2507).

Cement (MLH 464)

The agreement of the Cement Makers Federation (R77*), which provided for
common prices and aggregated rebates, was contested in the Court in 1961,
and the price restrictions were allowed to continue (see LR 2 RP 241 and the
report of the Director General of Fair Trading for 1974). The agreement was
still in operation in the late 1970s. Swann et al. (1973) describe the evolution
of the cement industry from the 1950s to the early 1970s. Collusion continued
in the industry throughout the period examined in this book.

Abrasives (MLH 469.1)

No agreements were registered, but PEP mentions the Abrasive Industries


Association as allegedly collusive.

Miscellaneous Building Materials and Mineral Products


(MLH 469.2)

Asbestos cement goods. According to MC, Asbestos and Certain Asbestos


Products (London: H.M.S.O., 1973), there was a price agreement in the in-
dustry before 1956. The agreement was abandoned after the introduction of
the 1956 Act and was not registered. The industry was also listed in both BT
and PEP as being subject to collusive pricing and other restrictions. The MC
report also stated that competition was generally effective after 1956. Hence
there was a change of competition regime in the asbestos cement industry.

Precast concrete goods. The register contains several agreements between


members of regional associations. They all contained price restrictions, and
some also provided for the reporting of inquiries and the allocation of work
between the parties; there were also agreements between associations to ob-
serve each other’s arrangements in their respective areas (R196, R197, R198,
R203, R268, R511, R512*, R521, R539, R540, R909, R910, R1036, R1037, R1242,
R1244, R1276, R1352, R1463, R1490, R1671, R1795, R2606). These agreements
covered a variety of precast concrete goods: paving slabs and curbs were
usually covered, and sometimes partition blocks, pipes and tubes, and other
products as well. All the agreements were abandoned in 1958–1960. Two
Survey of Collusive Agreements in British Industries 403

important national agreements ended in 1959: R305, which was operated by


the Federation of Building Blocks Manufacturers and contained prices for
building blocks, and R951*, which was operated by the British Concrete Pipe
Association and provided for minimum prices and the allocation of work for
concrete pipes, tubes, and related products in England and Wales. Overall,
the largest part of the industry was probably cartelized in the 1950s.
The state of competition in the 1960s and 1970s is less clear. A price
agreement by the North East Concrete Producers Association was referred to
the Court and was abandoned in 1970 (R3140*). R951* was referred to the
Court in 1965, although it had been formally canceled in 1959. Several other
collusive agreements in precast concrete goods were in operation at various
dates during the 1970s. The most important of these were two national
agreements in concrete pipes and manhole components, both of which were
discontinued in 1978 (R4556* and R4758*). Finally, according to MMC, Con-
crete Roofing Tiles (London: H.M.S.O., 1981), competition in concrete roofing
tiles was muted in the 1970s. The products affected by these arrangements
accounted for about 20–30% of total industry sales in the 1970s.

Ready-mixed concrete. The register contains a very large number of local


agreements in ready-mixed concrete that operated at various dates during
the 1960s and the 1970s (although, oddly, it contains no agreements for the
1950s). The vast majority of these were discovered during an official investi-
gation into the industry carried out in the 1970s. They were placed on the
register (and most were also referred to the Court), and were abandoned in
the late 1970s. Details can be found in the annual reports of the Director
General of Fair Trading for 1978 and 1979 and in MMC, Ready Mixed Concrete
(London: H.M.S.O., 1981). According to the MMC, however, the agreements
were of limited scope and were not always observed. Thus the state of com-
petition in this industry throughout the period examined in this book should
be regarded as ambiguous.

Lime and whiting. The register contains price agreements by the Southern
Lime Association (R1248), the South West of England Lime Association
(R2268), and the British Whiting Federation (R1259). The lime industry was
also listed in CD. All agreements were abandoned in 1959, leading to a
change of competition regime in this industry.

Roofing felts. A pricing agreement by the Built-up Roofing Council, whose


members were manufacturers and/or contractors of built-up bituminous felt
roofing materials, was canceled in 1959 (R841). The roofing felt industry was
also listed in PEP and BT as allegedly collusive; PEP reported complaints
that the Association of British Roofing Felt Manufacturers was a price-fixing
body. According to CD and BT, there were collective discrimination arrange-
ments for damp courses before 1956, at least in certain regions. The register
also contains two agreements between asphalt and felt roofing manufacturers
that suggest price-fixing and the allocation of work may have continued, at
404 Appendix A

least to some extent, during the 1960s: a regional agreement that was in op-
eration until 1967 (R2818), and an agreement by some of the largest firms in
the industry with respect to a particular tender for a government contract in
1966–1967 (R3106). This evidence suggests that it may be best to classify this
industry as ambiguous.

Manufactured bituminous asphalt and emulsions, coated roadstone, and


coated slag. The Asphalt Roads Association (R216) and the British Tar-
paviers Federation (R118) specified standard schedules of charges until 1958.
The Scottish Hot Road Binder Manufacturers and Spraying Association fixed
prices until 1962 (R895). Several other groups or regional associations of as-
phalt manufacturers and contractors abandoned their collusive agreements
between 1958 and 1961 (R440*, R952, R1192, R1252, R1628, R1934). An agree-
ment by the Mastic Asphalt Employers Federation contained prices for labor
and materials and conditions of contract (R578). Some of the price restric-
tions were dropped in 1961, while others continued until 1969. A price-fixing
agreement by the Scottish Road Emulsion Association ended in 1958 (R598),
but most of the member firms continued to participate in a collusive ar-
rangement for cold bituminous emulsion until 1967 (R3272). Bituminous
emulsion was also mentioned in both PEP and BT as a potentially collusive
industry, and road-surfacing materials in general were also included in the
BT lists of allegedly collusive products.
An official investigation into the road-surfacing materials industry carried
out in the 1970s revealed a large number of regional or local restrictive agree-
ments, which had been in operation since the early 1970s or earlier. These
agreements related to hot rolled asphalt and coated slag. They were placed
on the register (and most were also referred to the Court), and were aban-
doned in the late 1970s. See the 1979 report of the Director General of Fair
Trading for details. Their coverage and effectiveness are, however, not very
clear. All in all, the state of competition in this industry throughout the 1960s
and the 1970s must be regarded as ambiguous.

Other building materials and mineral products. Most other products of


MLH 469.2 were subject to collusion in the 1950s. In fact, the only important
products for which there is no evidence of any price agreements are gypsum,
plaster, and plasterboard. However, plasterboard was listed in CD, which
implies that it was subject to collective discrimination arrangements before
1956; according to BT, these probably involved restrictions on entry into
distribution. Gypsum was also listed in BT, probably for the same reason.
The register contains restrictive agreements by the Granite Kerb Confer-
ence (R386, R2226, R2227, R2228), the National Tile Fireplace Makers Asso-
ciation (R112), the Architectural Cornish Granite Association (R293, R294,
R295), the Slag Wool Association (R1249), and various associations of man-
ufacturers or engravers of memorials and related goods (R1520, R1543,
R1575*, R2105, R2485). Memorials and fireplaces were also listed in CD or BT
Survey of Collusive Agreements in British Industries 405

as being subject to collective discrimination arrangements. All agreements


were abandoned in 1959–1960, except for some restrictions of R1575* that
continued until 1965. R386 was replaced for some time by an information
agreement. Moreover, an agreement by the Mica Trade Association, whose
membership consisted of manufacturers, importers, and dealers, required
members to deal with one another only (R311); this ended in 1959. The mem-
bers of the Glazed Cement Manufacturers Association agreed on charges and
conditions of contract until 1964 (R559). A pricing agreement of the Plaster
Ventilators Manufacturers Association continued until 1968 (R3064*). Finally,
PEP mentions allegations of collusive pricing by the Association of Wood
Wool Manufacturers and the Association of Paving Manufacturers (pro-
ducers of paving flags and curbs).

Timber (MLH 471)

Sawn, planed, dressed, or further processed hardwood and softwood.


The register contains several price-fixing agreements by regional associations
of sawmillers, which must have covered more than 20% but less than 50% of
the total market (R86, R513, R517, R676, R1194, R937). They were all aban-
doned in 1958–1959.

Plywood and textured boards. No evidence of any restrictive agreements.

Wood chipboard. No evidence of any restrictive agreements.

Builders’ woodwork. The English Joinery Manufacturers Association rec-


ommended price increases until 1959 (R191). According to PEP, standard
types of joinery were under price control; special types were not. The Scottish
Joinery and Door Manufacturers Association also abandoned its pricing
agreement in 1959 (R946). Hence the industry probably experienced a change
of competition regime as a result of the 1956 Act.

Other products and work done. The Kiln Owners Association set a sched-
ule of charges for kiln drying of timber and plywood until 1958 (R520).

Furniture and Upholstery (MLH 472)

An agreement by the British Furniture Trade Confederation, whose mem-


bership consisted of both manufacturers and distributors, required that
members deal only with one another and provided for the maintenance of
retail prices (R1288*). This agreement covered products of MLH 472, MLH
473 and MLH 399.1. It was abandoned in 1963, but it does not seem to have
been very restrictive, especially in light of the very large membership of the
406 Appendix A

confederation. A similar agreement by the Scottish House Furnishers Asso-


ciation also ended in 1963 (R1329*). An agreement by the British Furniture
Manufacturers Federated Associations related mainly to settlement discounts
and other conditions of sale (R1454*); most of the restrictions were dropped
in 1963. Two other agreements on discounts and conditions of sale, by the
Association of Folding Furniture Makers (R1464*) and the National Associ-
ation of Manufacturers and Distributors of Office Equipment (R1790), were
canceled in 1963–1964. These agreements are unlikely to have restricted
competition between manufacturers to any significant degree. This is con-
firmed by PEP, where the furniture industry is regarded as competitive. Fi-
nally, the register also contains a pricing arrangement between the members
of the Chair Frames Association (R4860), which apparently continued until
the late 1970s. In any case, this is a minor product of the industry. All in all,
the evidence suggests that the furniture industry was essentially competitive.

Bedding and Soft Furnishings (MLH 473)

The National Bedding Federation, which was a member of the British Fur-
niture Trade Confederation, had an agreement similar to the agreement of
the latter (see above, under MLH 472). In particular, members were to deal
only with one another and distributors were to maintain the manufacturers’
retail prices (R1686*). A separate agreement of the federation provided for
quantity discounts to be allowed by manufacturers on mattresses sold directly
to hotels (R173*). Both agreements ended in 1962–1963. A price agreement
by the National Association of Upholstery Fibre Processors, covering uphol-
stery and mattress pads, was abandoned in 1958 (R671). This was a second-
ary product of the industry, however. Thus the bedding and soft furnishings
industry was largely competitive.

Shop and Office Fittings (MLH 474)

A price agreement by the Shopfront Moulding Manufacturers Association


was abandoned in 1959 (R463). The scope of this agreement seems to have
been limited, however: the membership of the association consisted of only
eleven firms, while the total number of firms with more than twenty-five
employees in MLH 474 was at least 200 in the late 1950s. The register also
contains an agreement by the National Association of Shopfitters, whose
membership was large (R467). Unfortunately, it is not clear what the restric-
tions were in this case and when they were abandoned. It seems that the
association originally specified profit margins for contracting work and
conditions of sale. But there is no information about the time period during
which the restrictions were in force: they may have continued until the 1970s
or they may have been abandoned sometime after 1956 or they may have
been ineffective ever since the agreement was registered. This industry is not
mentioned as collusive in any other data source. It is probably best to regard
it as ambiguous.
Survey of Collusive Agreements in British Industries 407

Wooden Containers and Baskets (MLH 475)

The Shive Manufacturers Association fixed prices until 1959 (R2304). This is
a minor product of the industry. Neither this agreement, nor some local price
agreements on packing cases, abandoned at various dates between 1957 and
1967, could have had any significant effect on competition for the industry as
a whole.

Miscellaneous Wood and Cork Manufactures (MLH 479)

Cork manufactures. No evidence of any restrictive agreements.

Wood manufactures. The register contains pricing agreements by the As-


sociation of Wood Block Manufacturers (R1053), the Wood Handle Manu-
facturers Council (R566), the Brush Wood Turners Association (R551), and
the Underwood Products Association, which consisted of producers of
chestnut fencing and related products (R1323). The first three agreements
ended in 1959, and the fourth appears to have continued until 1972. The
affected products accounted for more than 10% but less than 50% of this
product category. PEP also mentions allegations of collusion between man-
ufacturers of coffins.

Paper and Board (MLH 481)

Paper, uncoated or coated. A large number of price-fixing agreements by


various associations and groups of paper manufacturers, covering nearly all
types of paper, were registered. Most of these agreements covered specific
types and grades of paper, so it would be tedious to give details on each
here.7 Nearly all were abandoned in 1958–1960, but a few were abandoned
in 1962–1963 and some were replaced by information agreements that lasted
until the mid- or late 1960s. According to CD, parts of the industry were
subject to collective discrimination arrangements before 1956. In the case of
newsprint, where prices were under government regulation until 1956 and
were then fixed by the Association of Makers of Newsprint between 1956
and 1960 (R1328), there were also, until 1958, international agreements lim-
iting imports (R2433). Finally, two agreements by the British Paper and
Board Makers Association, relating to minimum buying prices of waste paper
from local authorities (R925*) and minimum and maximum buying prices of
waste paper from merchants (R71*), were contested in the Court and struck

7. They are the following: R497, R498, R499* (by the Association of Makers of Wood
Free Papers), R500, R501, R502, R503, R504, R505, R506, R507, R556* (by the Associa-
tion of Corrugated Paper Makers), R923, R978, R1237, R1238, R1239, R1240, R1241,
R1265, R1266, R1267, R1268, R1282, R1328, R1361, R1364, R1365, R1366, R1367, R1368,
R1370, R1456, R1501, R1735, R2224, R2635.
408 Appendix A

down in 1963 (LP 4 RP 1, LR 6 RP 161). To summarize, there was a change of


competition regime for all the principal products of the paper industry.
However, a 1967 official report on the price of newsprint made allegations
about the existence of an international cartel in this product, involving British
and foreign producers. See National Board of Prices and Incomes, Prices of
Standard Newsprint (London: H.M.S.O., 1967).

Board, uncoated or coated. The register contains pricing agreements for


intermittent board (R924), fourdrinier chipboard (R1238, R1370), and certain
types of coated board (R1267). They were all abandoned in 1960. The share
of these products in the total sales of board was probably between 10% and
50%. On the other hand, the agreements by the British Paper and Board
Makers Association on waste paper discussed above are unlikely to have had
any major effect on competition in the product market. Finally, BT reported
allegations about collective discrimination arrangements in hardboard and
fiber building board. According to PEP, however, there was no evidence that
sales of building board were regulated.

Cardboard Boxes, Cartons, and Fibre-Board Packing Cases


(MLH 482.1)

Paper boxes. No evidence of any significant restrictive agreements. The


British Paper Box Federation specified only conditions of sale (R1406).

Cartons. No evidence of any significant restrictive agreements. The British


Carton Association specified only conditions of sale (R1404).

Fibreboard packing cases. An agreement by the Fibreboard Packing Case


Manufacturers Association contained only a recommendation not to offer
discounts (R1230). The industry was basically competitive.

Other products. No evidence of any restrictive agreements.

Packaging Products of Paper and Associated Materials, Other Than


Board (MLH 482.2)

A price agreement by the British Paper Bag Federation fixed prices for paper
bags until 1965 (R1235). Moreover, the Royal Hand and Grocery Bags Asso-
ciation set prices for soda bags and sugar bags until 1960 (R1369). Paper bags
accounted for about 30% of total sales revenue in the whole of MLH 482.2
and for about 45% of total sales revenue of the product category ‘‘paper bags
and sacks’’ in the 1960s.
Survey of Collusive Agreements in British Industries 409

Manufactured Stationery (MLH 483)

A price agreement by the Envelope Makers and Manufacturing Stationers


Association, covering envelopes, account books, and personal stationery,
was abandoned in 1959 (R987). A price agreement by a group of producers
of various types of stationery for use by local authorities also ended in 1959
(R209). The products affected by these agreements accounted for some 30–
40% of industry sales revenue in the 1960s. According to the Lloyds’ report,
the Stationers Proprietary Articles Trade Association, whose membership
consisted of both manufacturers and distributors, practiced resale price
maintenance and set minimum distributors’ margins in the stationery indus-
try until 1956, but it did not regulate the manufacturers’ individual prices.
The stationery industry was also listed in CD.

Miscellaneous Manufactures of Paper and Board (MLH 484)

Wallpaper. The members of the Wallpaper Manufacturers and Employers


Association agreed upon the retail prices of wallpaper distributed by spe-
cialist merchants, but they did not regulate the prices of wallpaper sold
through retail shops (R201). According to CD and BT, they also engaged in
collective discrimination arrangements, including collective exclusive deal-
ing, until 1956. The pricing restrictions were terminated in late 1957. The
‘‘price-controlled range’’ accounted for about 50% of total sales of wallpaper
in the late 1950s, but this fraction declined in the 1960s. See MC, Report on the
Supply of Wallpaper (London: H.M.S.O., 1964).

Other products. The Association of Lace Paper Makers (R495), the Associ-
ation of Makers of Paper Serviettes (R496), the Society of Crepe Paper Makers
(R491), and the Federation of Paper Tube Manufacturers (R183) fixed prices
for their respective products until 1959. A price agreement by the Vulcanised
Fibre and Leatheroid Association ended in 1958 (R1076).

Printing, Publishing of Newspapers and Periodicals (MLH 485, 486)

Newspapers. An agreement between the Newspaper Proprietors Associa-


tion and the National Federation of Retail Newsagents, Booksellers and Sta-
tioners contained provisions that restricted the entry of new retailers into
newspaper distribution (R97*). This is probably the reason why the news-
paper industry was listed in CD. The agreement was contested in the Court
and was struck down in 1961 (LR 2 RP 453).
More significant with respect to competition between publishers was a
series of agreements between newspaper proprietors, including proprietors
of London daily and Sunday newspapers, relating to common retail price
increases and agreed trade discounts (R2001, R2002, R2718, R2719, R2720*,
410 Appendix A

R2721, R2722, R2805*, R2909, R2965*). Most of these agreements were dis-
continued in 1964–1966, although in a few cases agreements on price
increases were made as late as 1970 (R3590*). Several other regional agree-
ments provided for the prenotification of price increases and restrictions on
sales promotion; in fact, in most cases the firms involved were required not
to change the prices of their newspapers without prior discussion with the
other parties (R2611*, R2394–R2420). All these ended in 1960–1962. Other
agreements between publishers, mostly regional, comprised conditions of
sale (including a requirement not to sell newspapers before a certain time of
the day), sales promotion, or restrictions on the entry of retailers, and ended
at various dates between 1957 and 1964 (R1653*, R2003, R2004, R2299,
R2300, R2387, R2388, R2422, R2610, R2612). Finally, a series of agreements
between publishers regulating the distribution of newspapers and providing
for agreed trade discounts and conditions of sale were abandoned in the late
1960s or the early 1970s (R2385*, R2386, R2604, and many others similar in
form to those three).
In summary, the industry was probably collusive in the 1950s. It experi-
enced a change of competition regime as a result of the 1956 Act, although
the effect of the Act was probably not fully realized until the 1970s.

Periodicals. No evidence of any significant restrictive agreements affecting


publishers. Two agreements by the Periodical Proprietors Association,
abandoned in 1961–1962, related only to conditions of sale and restrictions
on the entry of retailers (R871, R875), which probably explains why the in-
dustry was listed in CD.

General Printing, Publishing, etc. (MLH 489)

Published books. No evidence of any significant restrictive agreements af-


fecting publishers. An agreement between publishers relating to the mainte-
nance of individual resale prices (R1586*, R1587*; see also CD and the Lloyds’
report) was contested in the Court in 1962 and was allowed to continue (LR 3
RP 246). This did not greatly affect competition between publishers, how-
ever, which, as the Court pointed out, was intense. The register also contains
agreements on discounts and conditions of sale for special kinds of books,
such as educational books, condensed books, and book club editions (R1650,
R2514, R2781). Most of the restrictions in these agreements were dropped in
the early 1960s, although some remained. In summary, this was a largely
competitive industry.

Other published products. An agreement by the Music Publishers Associ-


ation, comprising various pricing restrictions, continued until 1973 (R3641,
R3642, R3643). An agreement by the Diary Publishers Association, which
provided for maximum trade discounts, the maintenance of individual list
prices, and conditions of sale, ended in 1961 (R820, R1619, R2283, R2520,
Survey of Collusive Agreements in British Industries 411

R2608*). PEP also mentions the Greeting Card and Calendar Association as
potentially collusive.

General printing, etc. work. The register contains price-fixing agreements


by the Ceramic Printers Association (R70), the British Cinema and Theatre
Printers Association (R73), the British Embroidery Transfer Manufacturers
Association (R171), the Federation of Master Process Engravers (R278*), the
Photo-Litho Reproducers Association (R372), the National Association of
Engravers and Die-Stampers (R526), the Law Stationers Association (R1225),
the Association of Northern Master Electrotypers and Stereotypers (R1254),
the Society of Photo Printers (R1445), the Small Offset Association (R1929), the
London Trade Typesetters Association (R2444), and the Electrotyping and
Stereotyping Employers Federation (R2747*). In the case of the Fine Art
Trade Guild (R822), the arrangement originally provided for price-fixing,
collective exclusive dealing, and the collective enforcement of resale prices
(see CD). Many of these agreements covered preparatory or nonstandard
printing work. All were terminated in 1958–1959, except for R2747*, which
ended in 1965. A price agreement between the British Federation of Master
Printers and the Society of Parliamentary Agents covered only the printing
of parliamentary documents (R1038, R1333); it was canceled in 1960. The
effect of the 1956 Act on the printing industry as a whole should be seen as
ambiguous, although it caused a change of competition regime in specific
sections of the industry.

Rubber (MLH 491)

Tyres and tubes. The agreement of the Tyre Manufacturers Conference


provided for common prices for ‘‘replacement’’ tires, discussions about price
changes in the case of ‘‘original equipment’’ tires, collective exclusive deal-
ing, aggregated rebates, collective resale price maintenance, and restrictions
on sales promotion. There were also international agreements limiting im-
ports. See MRPC, Report on the Supply and Export of Pneumatic Tyres (London:
H.M.S.O., 1955). A modified version of this agreement was registered (R964*,
R966, R967, R1327) and then abandoned in 1961. It was replaced by an infor-
mation agreement, which was referred to the Court in 1965 and was found to
have amounted to the same effect as the former explicit pricing arrangement.
The information agreement ended in 1965.
The register also contains a series of agreements between the Tyre Manu-
facturers Conference and associations of distributors, which effectively re-
stricted entry into tire distribution (R963*, R965, R2272, R2273, R2274, R2436,
R2509*, R2632, R2633). These agreements were unsuccessfully contested in
the Court in 1963 (LR 3 RP 404). Finally, the MRPC report also mentions that
the Retread Manufacturers Association recommended prices to its members
for retreaded tires; the registered agreement of this association contained
conditions of sale only (R2828). All in all, the 1956 Act caused a change of
412 Appendix A

competition regime in the tire industry in the long run. According to Swann
et al. (1973), competition between manufacturers was intense in the early
1970s.

Rubber belting. No agreements were registered. However, the industry


was listed as allegedly collusive in both BT and PEP.

Rubber hose and tubing. No agreements were registered. However, the


industry was listed as allegedly collusive in PEP.

Other rubber goods and work done. Price agreements by the Rubber
Proofers Association (R914*), the Screw Stopper Makers Association (R1355),
and a group of producers of insulated, varnished, nonadhesive cambric or
silk cloth and tape (R710) were abandoned in 1957–1959.

Linoleum, Plastics Floor Covering, Leather Cloth, etc. (MLH 492)

The agreement of the Linoleum Manufacturers Association comprised prices,


some form of profit pooling, collective exclusive dealing, aggregated rebates,
and restrictions on media advertising. In addition, the association was a
party to an international collusive arrangement in linoleum. See MRPC, Re-
port on the Supply of Linoleum (London: H.M.S.O., 1956); the industry is also
listed in CD. A revised version of the agreement was registered (R729*), and
unsuccessfully contested in the Court in 1961 (LR 2 RP 395). Following the
Court’s decision, the registered agreement was replaced by a watered-down
version, which continued until 1963. A price agreement between manufac-
turers of felt base was canceled in 1965 (R2839*). Finally, a price agreement
between manufacturers of leather cloth ended in 1959 (R670). The products
affected by these agreements accounted for more than half of total industry
sales. Thus the 1956 Act caused a change of competition regime for this in-
dustry as a whole, as well as for most of its principal products.

Brushes and Brooms (MLH 493)

There is no evidence of any restrictive agreements in this industry.

Toys, Games, Children’s Carriages, and Sports Equipment


(MLH 494)

Toys and games. No evidence of any significant restrictive agreements.

Children’s carriages. No evidence of any significant restrictive agreements.


Survey of Collusive Agreements in British Industries 413

Sports equipment. The register contains a pricing agreement on cricket


and hockey balls, which ended in the late 1960s or early 1970s (R3048, R3098).
According to PEP, there were price-fixing arrangements in certain other
sports goods as well (this information comes from a trade association in the
industry). Tennis balls were perhaps a case in point. The share of all the
regulated products in total industry sales revenue was probably larger than
10%. According to CD, some sports goods were subject to collective dis-
crimination arrangements before 1956.

Miscellaneous Stationers’ Goods (MLH 495)

Pens and mechanical pencils. No evidence of any restrictive agreements


affecting competition between manufacturers. According to the Lloyds’ re-
port, the Stationers Proprietary Articles Trade Association practiced resale
price maintenance and set minimum distributors’ margins in the stationery
industry (until 1956), but did not regulate the manufacturers’ individual
prices. The stationery industry was also listed in CD.

Lead Pencils. The Pencil Makers Conference fixed the prices of cheaper
pencils, although not the prices of the more expensive varieties (R1426). This
may be seen as a minimum-price agreement. Hence the pencil industry
probably experienced a change of competition regime as a result of the 1956
Act.

Other stationers’ goods. No evidence of any restrictive agreements affect-


ing competition between manufacturers.

Plastics Products (MLH 496)

A price agreement between manufacturers of polyethylene tubing ended in


1962 (R272). On the other hand, the register contains evidence that manu-
facturers of PVC pressure pipes engaged in discussions about prices and
discounts until 1972 (R3167*, R3215, R3731*, R3732*, R3733*), while pro-
ducers of polyethylene pipes were parties to a collusive arrangement during
1974–1980 (R5072*). A restrictive agreement on plastic door furniture (knobs,
locks, latches, etc.) by the Plastics Hardware Association ended in 1959
(R302). A pricing arrangement between the members of the Flushing Cistern
Association continued unregistered until 1971 (R3671*), and was contested
in the Court (but struck down). An agreement by the Lampshade Manu-
facturers Association, abandoned in 1959, comprised only conditions of sale
(R360), and the same was the case for an agreement by a group of producers
of plastic moldings, which ended in 1972 (R1322). Finally, a pricing agree-
ment by a group of manufacturers of adhesive tapes was abandoned in 1965
(R2970*, R2962*). The products affected by all these agreements represented
less than 10% of MLH 496. Hence the 1956 Act had no significant effect on
the industry as a whole and on most of its principal products.
414 Appendix A

Miscellaneous Manufacturing Industries (MLH 499)

Musical instruments. A restrictive agreement by the Federation of Master


Organ Builders (R1180) ended in 1959. Organs accounted for less than 10%
of total sales of musical instruments. The register also contains an agreement
by the Association of Musical Instruments Industries, specifying maximum
trade discounts and conditions of sale (R3749). This agreement apparently
was abandoned in 1974. Thus there was no change of competition regime in
the musical instrument industry during the period examined in this book.

Other products. The Briar Pipe Trade Association set minimum prices until
1959 (R600). An agreement by a group of producers of peat moss litter, relat-
ing to prices and the allocation of work, ended in 1961 (R799). Finally, BT
reports allegations about a restrictive association in casein buttons, although
it does not specify the types of restrictions. This may refer to an exclusive
dealing arrangement between sellers of casein plastic and producers of
casein buttons, also reported in CD.
Appendix B: Datasets
Table B1

416
Data set for chapter 4
Industry Year C5 SS DS K/N K/L CHANGE
Clay, brick earth, marl, and shale 1958 0.726 13066 12745 0.51 4.91 1
1963 0.821 17827 15996 0.74 7.21 1
1968 0.910 32001 32001 1.25 11.72 1
1975 0.966 33165 26369 1.91 17.21 1
Iron ore and ironstone 1958 0.789 13502 10856 1.49 7.68 0
1963 0.884 12748 11475 1.88 11.26 0
White flour for breadmaking 1958 0.715 121751 123922 0.77 4.60 1
1963 0.792 113211 119913 1.14 6.45 1
1968 0.808 110439 110439 1.56 9.50 1
1975 0.895 135726 108612 2.29 13.20 1
Bread 1958 0.315 184559 238086 0.46 2.57 1
1963 0.714 214081 245906 0.58 2.47 1
1968 0.773 236006 236006 0.86 2.93 1
1975 0.823 207525 212485 0.92 3.42 1
Bacon and ham, cured and smoked 1958 0.308 90730 86486 0.21 1.43 1
1963 0.474 74194 78009 0.31 1.70 1
1968 0.569 67902 67902 0.55 2.44 1
1975 0.532 102384 74775 0.66 3.60 1

Appendix B
Sausages and sausage meat 1958 0.505 44575 44238 0.21 1.43 0
1963 0.522 52110 52719 0.31 1.70 0
1968 0.562 59218 59218 0.55 2.44 0
1975 0.565 62598 58139 0.66 3.60 0
Preserved fruit, other than marmalade and 1958 0.308 23915 19305 0.46 1.95 0

Data Sets
jams 1963 0.377 20428 19915 0.86 2.99 0
1968 0.457 23267 23267 1.28 4.28 0
1975 0.444 18855 18775 2.12 6.33 0
Vegetable and seed oils 1963 0.823 85828 84149 1.11 7.75 0
1968 0.843 80519 80519 1.36 10.85 0
1975 0.869 122108 86560 2.28 16.85 0
Fish and marine animal oils 1958 0.949 12364 5893 0.93 6.31 0
1963 0.928 14080 9922 1.11 7.75 0
1968 0.929 14129 14129 1.36 10.85 0
1975 0.913 20553 12343 2.28 16.85 0
Oilseed cake and meal, meat meal, 1963 0.804 27226 26755 1.11 7.75 0
bonemeal, etc. 1968 0.726 21508 21508 1.36 10.85 0
1975 0.817 34060 34840 2.28 16.85 0
Malt 1958 0.534 24538 17595 0.81 3.78 0
1963 0.551 30607 25993 1.36 5.31 0
1968 0.651 30347 30347 3.03 8.41 0
1975 0.724 62701 54227 7.20 16.76 0
Coke 1958 0.619 162360 154766 6.02 21.34 0
1963 0.653 153464 149435 9.73 28.18 0
Steel ingots 1958 0.842 55246 45286 7.54 11.35 0
1963 0.766 48954 45106 12.15 17.26 0
Steel blooms, billets, and slabs 1958 0.722 258111 223065 7.54 11.35 0
1963 0.696 203506 187510 12.15 17.26 0

417
Table B1 (continued)

418
Industry Year C5 SS DS K/N K/L CHANGE
Steel arches, etc. 1958 0.636 27664 23575 7.54 11.35 0
1963 0.692 21136 19474 12.15 17.26 0
Hot rolled coil, etc. 1958 0.688 135593 116048 7.54 11.35 0
1963 0.728 174121 160434 12.15 17.26 0
Steel plates, uncoated, 3 mm thick and 1958 0.789 159323 131141 7.54 11.35 0
over 1963 0.724 130471 120215 12.15 17.26 0
Steel plates and sheets, uncoated, under 3 1958 0.911 173290 145342 7.54 11.35 0
mm thick 1963 0.942 166268 153199 12.15 17.26 0
Wrought tubes, manipulated, fabricated, 1958 0.858 33584 32245 2.25 4.92 1
etc. 1963 0.733 30886 30007 3.21 7.29 1
1968 0.793 35437 35437 3.68 8.03 1
Pig iron 1958 0.547 125126 107090 0.61 3.57 0
1963 0.673 97637 89962 0.97 4.82 0
Iron castings 1958 0.250 225786 222694 0.61 3.57 1
1963 0.329 266582 259289 0.97 4.82 1
1968 0.422 272333 272333 1.21 5.65 1
Aluminium and aluminium alloys, and 1958 0.504 177881 162181 1.39 4.22 1
manufactures thereof 1963 0.542 218852 220888 1.55 5.02 1
1968 0.509 282354 282354 2.22 7.32 1

Appendix B
1975 0.541 299943 316371 4.24 15.30 1
Copper and manufactures thereof 1958 0.746 149102 265497 1.23 5.05 1

Data Sets
1963 0.741 179616 307212 1.35 5.80 1
1968 0.798 254943 254943 1.64 6.71 1
1975 0.830 150489 239120 1.94 7.76 1
Nickel and nickel alloys, and 1963 0.954 55743 66342 1.83 7.17 0
manufactures thereof 1968 0.970 63786 63786 2.13 9.81 0
Tin and tin alloys, and manufactures 1958 0.913 42598 60224 1.74 6.99 0
thereof 1963 0.893 36996 45224 1.83 7.17 0
1968 0.916 60136 60136 2.13 9.81 0
1975 0.978 41179 38656 2.69 15.48 0
Zinc and zinc alloys, and manufactures 1958 0.686 28233 35841 1.74 6.99 0
thereof 1963 0.690 37667 41810 1.83 7.17 0
1968 0.713 44085 44085 2.13 9.81 0
1975 0.734 29847 18562 2.69 15.48 0
Track-laying tractors 1958 0.902 8315 9536 1.29 3.43 0
1963 0.980 18749 18488 1.58 3.64 0
Lifts, escalators, and passenger conveyors 1958 0.625 12422 12452 0.29 1.41 1
1963 0.616 15513 14986 0.33 1.56 1
1968 0.641 20256 20256 0.41 1.91 1
1975 0.751 29538 28427 0.53 2.44 1
Mining machinery 1958 0.362 36573 36848 0.51 2.09 0
1963 0.422 43586 42979 0.56 2.38 0
1968 0.591 38306 38306 0.70 2.81 0
Space-heating, ventilating, and air- 1958 0.233 45635 45978 0.37 2.13 0

419
conditioning equipment (1958 S.I.C.) 1963 0.241 78532 77438 0.36 2.06 0
Table B1 (continued)

420
Industry Year C5 SS DS K/N K/L CHANGE
Fans and ventilating units, industrial (1968 1963 0.737 15938 15716 0.36 2.06 0
S.I.C.) 1968 0.604 19641 19641 0.41 2.41 0
1975 0.566 21960 20764 0.50 3.31 0
Bottling, packing, canning, packeting, and 1963 0.553 17937 17687 0.49 2.23 0
labeling machinery 1968 0.567 21958 21958 0.59 2.82 0
1975 0.592 22718 22118 0.71 3.84 0
Automatic slot machines 1963 0.903 2604 2567 0.49 2.23 0
1968 0.721 15305 15305 0.59 2.82 0
1975 0.724 8305 8173 0.71 3.84 0
Pulpmaking, papermaking, and 1958 0.840 10981 11063 0.42 2.07 0
boardmaking machinery 1963 0.938 22077 21770 0.49 2.23 0
1968 0.833 16208 16208 0.59 2.82 0
1975 0.825 13125 12916 0.71 3.84 0
Ordnance and small arms 1958 0.712 51490 51877 1.61 2.58 0
1963 0.625 40566 40001 2.42 3.45 0
1968 0.766 55543 55543 3.16 5.08 0
Bearings, other than ball and roller 1958 0.847 16338 16461 0.76 3.48 0
bearings, and bushes 1963 0.739 26617 26246 0.73 3.62 0
1968 0.866 26012 26012 0.92 3.95 0

Appendix B
1975 0.822 30512 28849 1.07 5.36 0
Gaskets and jointings 1963 0.762 8337 8221 0.73 3.62 0
1968 0.806 16709 16709 0.92 3.95 0
1975 0.904 11250 10636 1.07 5.36 0
Electric cables, other than for 1958 0.533 106228 114259 3.00 4.28 1

Data Sets
telecommunication 1963 0.680 137090 154163 2.84 4.68 1
1968 0.818 149872 149872 3.75 5.42 1
Electric light fittings, wiring and 1958 0.381 41734 37492 0.49 1.72 1
accessories, etc. 1963 0.380 61029 60668 0.63 2.03 1
1968 0.473 67898 67898 0.98 2.68 1
1975 0.470 84145 81542 1.43 4.13 1
Trailers and caravans 1958 0.370 24164 24469 3.47 6.46 1
1963 0.435 36093 36472 3.49 5.81 1
1968 0.442 71277 71277 3.71 6.49 1
1975 0.346 79799 73079 3.78 6.78 1
Hand tools and implements 1958 0.286 30391 32426 0.29 1.92 1
1963 0.333 38833 39194 0.35 2.34 1
1968 0.376 39526 39526 0.41 2.73 1
1975 0.522 45951 39840 0.55 3.62 1
Iron and steel wire 1958 0.561 73452 70269 0.60 3.42 1
1963 0.577 80919 78527 0.71 4.15 1
1968 0.573 82320 82320 0.95 5.27 1
1975 0.655 87046 68175 1.20 6.26 1
Manufactures of iron and steel wire 1958 0.417 59268 59338 0.60 3.42 1
1963 0.445 76581 77202 0.71 4.15 1
1968 0.512 86233 86233 0.95 5.27 1
1975 0.483 131816 122915 1.20 6.26 1
Wire of brass and other copper alloys, and 1958 0.690 11496 17529 0.60 3.42 1
manufactures thereof 1963 0.752 14570 18442 0.71 4.15 1

421
1968 0.829 18604 18604 0.95 5.27 1
Table B1 (continued)

422
Industry Year C5 SS DS K/N K/L CHANGE
Copper wire and manufactures thereof 1958 0.698 41459 72366 0.60 3.42 1
1963 0.724 23503 43499 0.71 4.15 1
1968 0.869 38609 38609 0.95 5.27 1
Cans, metal boxes, and other small metal 1958 0.869 91347 80517 0.84 2.98 0
containers 1963 0.904 109910 104299 1.47 4.07 0
1968 0.908 126711 126711 2.24 6.04 0
Precious metals, refined 1963 0.954 83585 105953 0.36 2.56 0
1968 0.965 138771 138771 0.68 4.10 0
Jewellery and plate (1958 S.I.C.) 1958 0.218 25559 22731 0.30 2.23 0
1963 0.263 34946 29945 0.36 2.56 0
Jewellery and plate (1968 S.I.C.) 1963 0.336 23130 19820 0.36 2.56 0
1968 0.495 27497 27497 0.68 4.10 0
1975 0.441 63290 49942 0.82 4.95 0
Metal furniture 1958 0.241 42492 41095 0.29 2.47 0
1963 0.276 57570 57230 0.27 1.87 0
1968 0.225 66030 66030 0.33 2.24 0
1975 0.249 64974 63022 0.42 3.23 0
Drop forgings of steel and iron, and steel 1963 0.583 103497 89132 0.41 2.40 1
stampings and pressings 1968 0.612 104139 104139 0.50 2.99 1

Appendix B
1975 0.650 115994 95893 0.58 3.72 1
Metal hollowware, domestic and 1958 0.252 61627 68700 0.33 2.14 1
industrial 1963 0.335 66378 71636 0.41 2.40 1
1968 0.327 67972 67972 0.50 2.99 1
1975 0.412 60195 54461 0.58 3.72 1
Metal windows, metal door frames, etc. 1958 0.705 26531 27460 0.33 2.14 1

Data Sets
1963 0.583 30510 30294 0.41 2.40 1
1968 0.610 33217 33217 0.50 2.99 1
1975 0.509 52420 34309 0.58 3.72 1
Safes, locks, latches, keys, etc. 1958 0.687 9517 12377 0.33 2.14 1
1963 0.764 15224 18296 0.41 2.40 1
1968 0.743 18948 18948 0.50 2.99 1
1975 0.753 21861 19916 0.58 3.72 1
Single yarn of cotton, glass fibres, and 1958 0.319 188803 159023 0.74 3.39 1
man-made fibres 1963 0.372 134557 132227 1.22 5.22 1
1968 0.503 123935 123935 1.81 7.46 1
1975 0.756 82616 78395 2.58 11.05 1
Finished thread for sewing, embroidery, 1958 0.851 26988 22009 0.74 3.39 0
etc. 1963 0.818 23737 22278 1.22 5.22 0
1968 0.879 27491 27491 1.81 7.46 0
1975 0.879 27080 25558 2.58 11.05 0
Woven cotton cloth in the loom state 1958 0.116 157737 145988 0.46 3.25 0
1963 0.193 83641 86041 0.75 4.80 0
1968 0.313 52050 52050 1.08 7.22 0
1975 0.437 27205 25111 1.44 9.85 0
Woven cloth of man-made fibres in the 1958 0.211 99105 78882 0.46 3.25 1
loom state 1963 0.358 91010 81405 0.75 4.80 1
1968 0.519 84540 84540 1.08 7.22 1
1975 0.636 59521 60036 1.44 9.85 1

423
Table B1 (continued)

424
Industry Year C5 SS DS K/N K/L CHANGE
Wool tops or slubbings 1958 0.301 121031 96334 0.39 2.49 0
1963 0.340 120401 86844 0.50 2.98 0
1968 0.547 86161 86161 0.73 4.37 0
1975 0.642 44690 52171 1.06 6.97 0
Yarn of wool, hair, and man-made fibres 1958 0.237 52506 46333 0.39 2.49 0
spun on the woollen system 1963 0.230 72396 65786 0.50 2.98 0
1968 0.339 68272 68272 0.73 4.37 0
1975 0.370 59914 63038 1.06 6.97 0
Woven worsted fabrics 1958 0.173 109476 100736 0.39 2.49 0
1963 0.267 100029 93473 0.50 2.98 0
1968 0.311 79842 79842 0.73 4.37 0
1975 0.438 50733 50160 1.06 6.97 0
Woven woollen fabrics 1958 0.177 103410 99048 0.39 2.49 0
1963 0.151 88506 86576 0.50 2.98 0
1968 0.240 79218 79218 0.73 4.37 0
1975 0.268 44284 43274 1.06 6.97 0
Jute yarn, cloth, and other manufactures of 1958 0.508 36494 42969 0.94 4.09 1
jute 1963 0.592 40097 43219 1.07 4.01 1
1968 0.689 38728 38728 1.24 4.97 1
1975 0.734 12990 15466 2.27 9.30 1

Appendix B
Rope, twine, net, and manufactures 1958 0.477 29692 26886 0.34 2.30 1
thereof 1963 0.648 33161 27166 0.45 2.86 1
1968 0.732 24197 24197 0.77 4.24 1
1975 0.821 21168 17356 1.30 6.82 1
Knitted, netted, or crocheted goods: 1958 0.256 40530 31566 0.60 4.26 0

Data Sets
underwear, shirts, and nightwear 1963 0.395 46060 37270 0.68 4.23 0
1968 0.531 48760 48760 0.86 4.92 0
1975 0.504 41022 46169 0.99 5.81 0
Knitted, netted, or crocheted goods: other 1958 0.150 88309 93221 0.60 4.26 0
clothing 1963 0.212 113456 105789 0.68 4.23 0
1968 0.332 139594 139594 0.86 4.92 0
1975 0.402 125522 163748 0.99 5.81 0
Woven carpets, carpeting, and carpet floor 1958 0.412 80749 77717 0.83 2.96 1
rugs 1963 0.423 84798 80025 0.91 3.16 1
1968 0.575 81014 81014 1.24 3.80 1
Bedding other than wool blankets 1963 0.446 28247 27350 0.08 0.76 0
1968 0.513 37014 37014 0.14 1.30 0
1975 0.555 49565 49447 0.24 2.37 0
Other made-up household textiles 1963 0.271 28631 27722 0.08 0.76 0
1968 0.266 26181 26181 0.14 1.30 0
1975 0.474 25377 21230 0.24 2.37 0
Canvas goods and sacks 1958 0.252 33245 26594 0.10 0.94 1
1963 0.280 27742 24382 0.12 1.15 1
1968 0.249 21177 21177 0.14 1.43 1
1975 0.313 17515 16040 0.14 1.73 1
Asbestos manufactures other than asbestos 1958 0.774 41279 37379 1.81 3.28 1
cement goods 1963 0.824 49883 47653 2.12 3.59 1
1968 0.862 58645 58645 2.89 4.45 1

425
1975 0.867 52739 60991 4.33 6.23 1
Table B1 (continued)

426
Industry Year C5 SS DS K/N K/L CHANGE
Leather, undressed 1958 0.382 24405 25969 0.20 1.98 0
1963 0.453 16743 17306 0.24 2.34 0
1968 0.556 13845 13845 0.36 3.28 0
1975 0.763 11320 12061 0.49 4.50 0
Leather, dressed 1963 0.225 63802 66350 0.24 2.34 0
1968 0.217 75429 75429 0.36 3.28 0
1975 0.339 62049 70324 0.49 4.50 0
Fellmongery 1958 0.461 8928 7496 0.20 1.98 0
1963 0.427 16260 11909 0.24 2.34 0
1968 0.422 10523 10523 0.36 3.28 0
1975 0.506 8753 9210 0.49 4.50 0
Manufactures of leather or leather 1958 0.213 19570 20523 0.04 0.42 0
substitutes 1963 0.208 26438 26747 0.06 0.58 0
1968 0.203 28918 28918 0.07 0.70 0
1975 0.245 31140 32183 0.09 1.01 0
Skins and furs, dressed, etc., and 1963 0.377 11243 10495 0.10 0.82 0
manufactures thereof 1968 0.495 10252 10252 0.14 1.13 0
1975 0.635 7938 9554 0.16 1.59 0
Weatherproof outerwear 1963 0.223 50884 45994 0.08 0.74 0

Appendix B
1968 0.306 42361 42361 0.10 0.93 0
1975 0.298 40485 50254 0.12 1.08 0
Men’s and boys’ tailored outerwear 1958 0.350 148333 129315 0.07 0.34 0

Data Sets
1963 0.380 158776 146296 0.09 0.47 0
1968 0.377 158184 158184 0.13 0.67 0
1975 0.283 144680 168807 0.17 0.99 0
Women’s and girls’ tailored outerwear 1958 0.152 82374 75103 0.05 0.53 0
1963 0.167 86343 82575 0.06 0.59 0
1968 0.201 85066 85066 0.09 0.78 0
1975 0.177 99846 123790 0.10 0.87 0
Industrial and heavy overalls, aprons, and 1958 0.241 18897 18872 0.06 0.47 0
jeans 1963 0.261 22866 22522 0.08 0.65 0
1968 0.274 22717 22717 0.12 0.91 0
1975 0.326 33242 33321 0.14 1.11 0
Women’s and girls’ light outerwear 1958 0.082 83327 75540 0.03 0.40 0
1963 0.129 76555 72727 0.05 0.52 0
1968 0.166 95702 95702 0.08 0.74 0
1975 0.160 133928 187785 0.09 0.92 0
Lingerie other than corsetry and brassieres 1958 0.202 35242 33001 0.03 0.40 0
1963 0.265 34980 33304 0.05 0.52 0
1968 0.414 37978 37978 0.08 0.74 0
1975 0.453 30193 38132 0.09 0.92 0
Infants’ wear 1958 0.177 16519 15026 0.03 0.40 0
1963 0.153 17190 16046 0.05 0.52 0
1968 0.297 20488 20488 0.08 0.74 0
1975 0.288 17143 25685 0.09 0.92 0

427
Table B1 (continued)

428
Industry Year C5 SS DS K/N K/L CHANGE
Gloves, other than sport, knitted, and 1958 0.284 11629 11687 0.06 0.67 0
rubber gloves 1963 0.327 12894 12642 0.06 0.73 0
1968 0.414 10734 10734 0.08 0.92 0
1975 0.485 9125 9222 0.13 1.47 0
Tiles, other than of precast concrete and 1958 0.648 12642 12407 0.40 1.59 0
brick earth 1963 0.702 16120 15331 0.54 1.98 0
1968 0.936 17562 17562 0.73 2.48 0
1975 0.932 17522 18289 0.94 2.97 0
Sanitary ware of earthenware 1958 0.641 6972 6327 0.40 1.59 1
1963 0.663 11399 11138 0.54 1.98 1
1968 0.951 12733 12733 0.73 2.48 1
1975 0.939 14665 15067 0.94 2.97 1
China, earthenware, etc., domestic and 1958 0.274 37170 43571 0.40 1.59 1
ornamental 1963 0.317 40657 43386 0.54 1.98 1
1968 0.484 46653 46653 0.73 2.48 1
1975 0.576 72952 69921 0.94 2.97 1
Electrical ware of porcelain, earthenware, 1958 0.769 8289 8289 0.40 1.59 1
or stoneware 1963 0.808 11521 11092 0.54 1.98 1
1968 0.857 12014 12014 0.73 2.48 1
1975 0.754 11610 10957 0.94 2.97 1

Appendix B
Glass containers 1958 0.632 51899 47427 0.83 2.65 1
1963 0.696 60251 60456 1.41 3.68 1
1968 0.873 70260 70260 2.13 5.45 1
1975 0.876 84865 77991 2.89 8.70 1
Cement 1958 0.855 77393 71133 2.71 10.58 0

Data Sets
1963 0.890 90038 84136 3.79 14.40 0
1968 0.890 101579 101579 6.26 23.88 0
1975 0.930 126930 111057 6.83 25.60 0
Asbestos cement goods 1958 0.983 19849 19923 0.46 3.39 1
1963 0.942 24842 25524 0.63 4.58 1
1968 0.986 24510 24510 0.81 6.12 1
1975 0.982 17075 15740 1.19 8.86 1
Plywood and veneers 1963 0.336 17458 20244 0.16 1.32 0
1968 0.444 35998 35998 0.18 1.52 0
Textured boards and wood chipboard 1963 0.555 8768 8538 0.16 1.32 0
1968 0.544 19395 19395 0.18 1.52 0
Builders’ woodwork and prefabricated 1958 0.192 51985 58608 0.13 1.17 1
building structures of timber 1963 0.184 83526 89132 0.16 1.32 1
1968 0.240 125719 125719 0.18 1.52 1
1975 0.329 126038 125619 0.27 2.66 1
Office, school, and other furniture 1958 0.225 31472 31313 0.10 0.78 0
1963 0.243 37553 37858 0.12 0.93 0
1968 0.257 40945 40945 0.19 1.38 0
1975 0.284 43702 38301 0.33 2.36 0
Wooden containers and baskets 1958 0.116 33083 32711 0.09 0.91 0
1963 0.162 36207 36717 0.10 1.05 0
1968 0.176 36502 36502 0.13 1.33 0
1975 0.217 33061 30625 0.15 1.80 0

429
Table B1 (continued)

430
Industry Year C5 SS DS K/N K/L CHANGE
Paper, uncoated 1958 0.526 250871 218468 2.74 8.72 1
1963 0.566 261650 260101 3.23 9.61 1
1968 0.548 262225 262225 4.91 13.74 1
Paper and board, coated 1958 0.407 39023 31856 2.74 8.72 1
1963 0.536 52930 46615 3.23 9.61 1
1968 0.632 62511 62511 4.91 13.74 1
Paper and board, oiled, waxed, and other 1958 0.519 18681 16651 2.74 8.72 1
waterproof wrappings 1963 0.603 16882 16401 3.23 9.61 1
1968 0.647 17131 17131 4.91 13.74 1
Boxes and cartons of paper and cardboard 1958 0.314 78812 81573 0.49 3.21 0
1963 0.303 98159 95149 0.62 3.64 0
1968 0.387 113954 113954 0.81 4.27 0
1975 0.385 142961 116943 1.10 6.30 0
Fibreboard packing cases 1958 0.736 61670 52442 0.49 3.21 0
1963 0.655 91999 84586 0.62 3.64 0
1968 0.675 119657 119657 0.81 4.27 0
1975 0.631 149089 114461 1.10 6.30 0
Rubber cellular products 1958 0.863 12392 10971 1.27 3.80 0
1963 0.774 11688 10815 1.54 4.06 0

Appendix B
Carbons and other office machinery 1958 0.633 10662 9709 0.29 1.63 0
requisites 1963 0.642 14472 13871 0.43 2.33 0
1968 0.749 17506 17506 0.56 3.41 0
1975 0.792 17211 18894 0.81 5.56 0
Components, accessories, and 1963 0.120 48989 37990 0.33 2.24 0

Data Sets
semimanufactured goods of plastics 1968 0.164 84334 84334 0.55 3.60 0
1975 0.136 113173 125688 0.82 5.96 0
Domestic, catering, and furnishing goods 1963 0.235 19127 14833 0.33 2.24 0
of plastics 1968 0.299 31250 31250 0.55 3.60 0
1975 0.254 45655 52945 0.82 5.96 0
Packaging materials of plastics 1963 0.443 37290 28918 0.33 2.24 0
1968 0.411 75053 75053 0.55 3.60 0
1975 0.302 116905 139264 0.82 5.96 0
Musical instruments 1958 0.454 5356 5929 0.16 1.38 0
1963 0.501 6643 6642 0.21 1.63 0
Notes: C5 is the five-firm sales concentration ratio.
SS is sales revenue (in £1000) deflated by the general producer price index.
DS is sales revenue (in £1000) deflated by industry-specific producer price indices.
K is the value of capital stock of the corresponding MLH industry in 1980 prices (in £ million).
L is employment of the corresponding MLH industry (in 1000).
N is the number of plants with at least 25 employees of the corresponding MLH industry.
CHANGE takes the value 1 for an industry with a change in competition regime and 0 otherwise.

431
Table B2

432
Data set for the concentration regressions of chapter 5

Industry Year C5 SS DS K=N K=L TVCLASS ADTYPE CHANGE


Flour, other than white 1958 0.639 36139 33254 0.77 4.60 0.55 1 1
flour for breadmaking 1963 0.716 42927 41191 1.14 6.45 0.55 1 1
1968 0.672 41609 41609 1.56 9.50 0.85 1 1
1975 0.816 44699 37173 2.29 13.20 0.85 1 1
Cereal breakfast foods 1963 0.977 37469 34156 1.14 6.45 0.85 2 0
1968 0.935 43215 43215 1.56 9.50 0.85 2 0
1975 0.914 55558 51992 2.29 13.20 0.85 2 0
Biscuits for human 1958 0.479 136106 127902 1.47 2.63 0.55 2 1
consumption 1963 0.655 140957 137413 2.19 3.69 0.55 2 1
1968 0.710 146946 146946 3.13 4.30 0.85 2 1
1975 0.789 169005 155520 4.27 4.92 0.85 2 1
Fish and fish products, 1958 0.932 9978 11186 0.21 1.43 0.85 1 0
quick-frozen 1963 0.917 25958 24769 0.31 1.70 0.85 1 0
1968 0.911 35375 35375 0.55 2.44 0.85 1 0
1975 0.781 73511 56958 0.66 3.60 0.85 1 0
Condensed milk 1958 0.909 24067 20038 0.62 3.65 0.25 2 1
1963 0.934 25072 21001 0.74 4.74 0.25 2 1
1968 0.944 23911 23911 0.92 5.81 0.85 2 1
1975 0.889 23228 20122 1.35 8.27 0.85 2 1

Appendix B
Milk powder 1958 0.737 14686 12692 0.62 3.65 0.25 1 1
1963 0.889 15051 13633 0.74 4.74 0.55 1 1
1968 0.847 18692 18692 0.92 5.81 0.85 1 1
1975 0.753 38195 23339 1.35 8.27 0.85 1 1
Ice cream and ice lollies 1958 0.942 22392 18076 0.62 3.65 0.55 2 0

Data Sets
1963 0.931 14079 15895 0.74 4.74 0.85 2 0
1968 0.912 19276 19276 0.92 5.81 0.85 2 0
1975 0.909 49483 30115 1.35 8.27 0.55 2 0
Cocoa products 1958 0.732 176041 163309 0.84 2.42 0.55 2 1
1963 0.823 171639 166754 1.40 3.40 0.85 2 1
1968 0.834 180991 180991 2.22 4.64 0.85 2 1
1975 0.836 218224 184031 3.27 6.72 0.85 2 1
Marmalade and jams 1958 0.629 36118 34021 0.46 1.95 0.55 2 0
1963 0.729 34564 33391 0.86 2.99 0.55 2 0
1968 0.756 31488 31488 1.28 4.28 0.55 2 0
1975 0.689 30137 25043 2.12 6.33 0.85 2 0
Vegetables, etc., preserved 1958 0.649 45235 35897 0.46 1.95 0.55 1 0
(1958 S.I.C.) 1963 0.672 47825 42912 0.86 2.99 0.55 1 0
Vegetables, etc., preserved 1963 0.653 65177 58482 0.86 2.99 0.55 1 0
(1968 S.I.C.) 1968 0.667 73798 73798 1.28 4.28 0.85 1 0
1975 0.703 101902 95830 2.12 6.33 0.85 1 0
Vegetables, quick-frozen 1958 0.986 7340 8287 0.46 1.95 0.85 1 0
1963 0.933 24913 23826 0.86 2.99 0.85 1 0
1968 0.971 35911 35911 1.28 4.28 0.85 1 0
1975 0.817 58919 62240 2.12 6.33 0.85 1 0
Pickles, sauces, and relishes 1963 0.680 28206 27342 0.86 2.99 0.25 2 0
1968 0.715 30293 30293 1.28 4.28 0.55 2 0
1975 0.679 41689 42479 2.12 6.33 0.55 2 0

433
Table B2 (continued)

434
Industry Year C5 SS DS K=N K=L TVCLASS ADTYPE CHANGE
Soups 1958 0.911 25732 21650 0.46 1.95 0.55 2 0
1963 0.925 39165 34385 0.86 2.99 0.55 2 0
1968 0.904 39560 39560 1.28 4.28 0.85 2 0
1975 0.865 44952 43104 2.12 6.33 0.85 2 0
Dog and cat foods 1958 0.861 24539 20935 0.42 2.77 0.55 2 0
1963 0.943 38669 40458 0.59 3.88 0.85 2 0
1968 0.941 58155 58155 0.89 6.16 0.85 2 0
1975 0.944 74677 78807 1.34 9.66 0.85 2 0
Margarine 1958 0.881 50183 36611 1.46 7.58 0.55 2 0
1963 0.928 42188 36814 1.91 9.57 0.85 2 0
1968 0.938 31784 31784 2.89 10.59 0.85 2 0
Compound fat 1958 0.833 22975 15888 1.46 7.58 0.55 1 0
1963 0.848 21280 19094 1.91 9.57 0.55 1 0
1968 0.826 16498 16498 2.89 10.59 0.55 1 0
1975 0.884 27612 19640 3.78 13.51 0.55 1 0
Coffee, and coffee and 1958 0.979 22897 30630 0.63 3.82 0.55 2 0
chicory extracts and 1963 0.984 31801 34494 0.96 5.27 0.55 2 0
essences
1968 0.937 46642 46642 1.67 8.07 0.85 2 0
1975 0.927 51875 62435 2.51 9.78 0.85 2 0

Appendix B
Alcoholic cider, perry, apple 1958 0.812 20540 21770 0.67 3.46 0.55 2 0
pectin, and British wines 1963 0.972 20075 22784 1.20 4.86 0.55 2 0
1968 0.916 30004 30004 1.39 5.68 0.55 2 0
1975 0.902 46976 55645 2.75 7.55 0.55 2 0
Cigarettes 1963 0.997 1198822 1258907 4.46 4.84 0.55 2 0

Data Sets
1968 0.999 1236281 1236281 7.47 7.69 0.00 2 0
Other manufactured tobacco 1963 0.980 145500 158625 4.46 4.84 0.55 2 0
1968 0.981 129145 129145 7.47 7.69 0.25 2 0
1975 0.988 94571 119509 12.36 10.25 0.55 2 0
Lubricating oils and greases 1963 0.695 86247 82575 0.80 6.01 0.25 1 0
1968 0.716 93079 93079 1.28 10.93 0.25 1 0
1975 0.740 138734 135633 1.68 12.95 0.55 1 0
Pharmaceutical preparations 1958 0.286 119871 89003 1.33 4.10 0.55 2 0
1963 0.287 168815 138163 1.68 4.93 0.55 2 0
1968 0.349 221326 221326 2.67 7.55 0.55 2 0
1975 0.393 341185 474269 4.73 11.40 0.55 2 0
Hair preparations 1958 0.670 16860 18822 0.41 2.02 0.55 2 0
1963 0.542 26510 27535 0.48 2.03 0.55 2 0
1968 0.570 31731 31731 0.92 2.96 0.55 2 0
1975 0.607 46032 60769 1.50 5.06 0.55 2 0
Dental preparations 1963 0.944 13587 12347 0.48 2.03 0.85 2 0
1968 0.874 14450 14450 0.92 2.96 0.85 2 0
1975 0.803 21586 24625 1.50 5.06 0.85 2 0
Other toilet preparations 1958 0.420 33779 37708 0.41 2.02 0.25 2 0
1963 0.373 50961 52931 0.48 2.03 0.25 2 0
1968 0.405 70585 70585 0.92 2.96 0.25 2 0
1975 0.421 99401 111633 1.50 5.06 0.55 2 0

435
Table B2 (continued)

436
Industry Year C5 SS DS K=N K=L TVCLASS ADTYPE CHANGE
Soap 1958 0.832 59413 60862 2.27 7.35 0.55 2 0
1963 0.809 53272 53019 3.53 9.14 0.85 2 0
1968 0.801 48123 48123 4.44 14.43 0.85 2 0
1975 0.764 52683 49235 4.34 14.57 0.85 2 0
Finished detergents 1958 0.904 45542 36105 2.27 7.35 0.55 2 0
1963 0.845 55501 48629 3.53 9.14 0.85 2 0
1968 0.799 61265 61265 4.44 14.43 0.85 2 0
1975 0.828 90114 99083 4.34 14.57 0.85 2 0
Polishes 1958 0.671 17337 18133 0.36 2.20 0.25 2 0
1963 0.796 17901 18436 0.47 3.04 0.25 2 0
1968 0.809 17249 17249 0.85 4.29 0.55 2 0
1975 0.849 14138 18195 1.31 7.31 0.85 2 0
Pesticides, disinfectants, and 1958 0.424 29238 22616 0.41 2.96 0.55 2 0
household deodorisers 1963 0.520 34655 31491 0.68 4.56 0.55 2 0
1968 0.534 44697 44697 1.28 7.49 0.55 2 0
1975 0.533 73971 80699 2.57 13.13 0.55 2 0
Powered industrial trucks 1958 0.698 9424 10001 0.29 1.41 0.00 1 0
and industrial tractors 1963 0.764 16297 15212 0.33 1.56 0.00 1 0
1968 0.555 42182 42182 0.41 1.91 0.00 1 0

Appendix B
1975 0.607 71547 66984 0.53 2.44 0.00 1 0
Lawn mowers 1963 0.940 9840 9703 0.49 2.23 0.00 1 0
1968 0.882 18368 18368 0.59 2.82 0.00 1 0
1975 0.852 19401 19091 0.71 3.84 0.00 1 0
Photographic and cinemato- 1963 0.706 14917 15978 0.54 3.10 0.25 2 0

Data Sets
graphic apparatus and 1968 0.843 38812 38812 1.02 2.87 0.25 2 0
appliances, and document
1975 0.787 54097 63083 1.78 7.28 0.25 2 0
copying equipment
Television receiving sets 1958 0.528 78964 81098 0.53 1.10 0.00 1 0
1963 0.816 66324 72215 0.55 1.00 0.00 1 0
1968 0.927 86103 86103 1.01 1.98 0.00 1 0
1975 0.850 135769 201970 1.22 2.80 0.25 1 0
Radio receiving sets 1963 0.579 23118 25140 0.55 1.00 0.00 1 0
1968 0.747 11556 11556 1.01 1.98 0.00 1 0
1975 0.835 4939 7389 1.22 2.80 0.25 1 0
Cooking apparatus and 1958 0.651 20221 16031 1.18 2.55 0.00 2 1
appliances, electric 1963 0.600 26332 22655 1.18 2.27 0.25 2 1
1968 0.859 40547 40547 1.43 3.00 0.25 2 1
1975 0.791 41854 52480 2.14 4.34 0.25 2 1
Washing machines, 1958 0.764 34924 25410 1.18 2.55 0.25 2 1
electrically operated 1963 0.852 55575 44969 1.18 2.27 0.25 2 1
1968 0.869 36915 36915 1.43 3.00 0.25 2 1
1975 0.981 44509 52813 2.23 4.79 0.25 2 1
Batteries and accumulators 1958 0.782 42602 48030 0.49 1.72 0.25 1 1
1963 0.787 52821 56933 0.63 2.03 0.25 1 1
1968 0.841 66757 66757 0.98 2.68 0.55 1 1
1975 0.908 74751 76167 1.43 4.13 0.55 1 1
Electric lamps 1958 0.744 23666 18932 0.49 1.72 0.25 1 1
1963 0.717 35084 29795 0.63 2.03 0.25 1 1

437
1968 0.866 40308 40308 0.98 2.68 0.25 1 1
1975 0.941 35415 41413 1.43 4.13 0.25 1 1
Table B2 (continued)

438
Industry Year C5 SS DS K=N K=L TVCLASS ADTYPE CHANGE
Cars 1958 0.901 521809 471432 3.47 6.46 0.00 1 0
1963 0.912 752514 685258 3.49 5.81 0.00 1 0
1968 0.992 853240 853240 3.71 6.49 0.00 1 0
1975 0.982 690957 648622 3.78 6.78 0.00 1 0
Cutlery 1958 0.658 22897 20524 0.24 1.75 0.55 2 0
1963 0.669 29222 26752 0.39 2.20 0.55 2 0
1968 0.709 33214 33214 0.56 3.42 0.85 2 0
1975 0.560 25796 27019 0.60 4.41 0.85 2 0
Knitted, netted, or crocheted 1958 0.214 81626 55153 0.60 4.26 0.25 1 0
goods: hosiery 1963 0.201 82066 66593 0.68 4.23 0.25 1 0
1968 0.433 91573 91573 0.86 4.92 0.25 1 0
1975 0.455 58546 91096 0.99 5.81 0.25 1 0
Tufted carpets, carpeting, 1958 0.720 8129 7823 0.83 2.96 0.25 1 0
and carpet floor rugs 1963 0.507 24855 23456 0.91 3.16 0.25 1 0
1968 0.518 61323 61323 1.24 3.80 0.25 1 0
1975 0.450 101350 131887 1.93 6.23 0.25 1 0
Men’s and boys’ shirts, 1958 0.191 50765 50194 0.06 0.47 0.55 1 0
underwear, and nightwear 1963 0.254 54607 52636 0.08 0.65 0.25 1 0
1968 0.355 54724 54724 0.12 0.91 0.25 1 0
1975 0.400 51410 56144 0.14 1.11 0.25 1 0

Appendix B
Corsets and brassieres 1958 0.308 31864 31902 0.14 0.85 0.25 2 0
1963 0.380 40321 38997 0.13 0.91 0.00 2 0
1968 0.590 40537 40537 0.15 1.06 0.55 2 0
1975 0.632 32675 39991 0.17 1.32 0.85 2 0
Men’s and boys’ footwear 1958 0.222 65714 61826 0.11 0.67 0.25 1 0

Data Sets
(1958 S.I.C.) 1963 0.303 72032 67606 0.15 0.87 0.25 1 0
Women’s and girls’ 1958 0.239 90546 85189 0.11 0.67 0.25 1 0
footwear (1958 S.I.C.) 1963 0.289 98844 92770 0.15 0.87 0.25 1 0
Footwear, other than rubber 1963 0.242 183223 171964 0.15 0.87 0.25 1 0
footwear (1963 S.I.C.) 1968 0.316 175642 175642 0.20 1.15 0.55 1 0
1975 0.400 160203 180788 0.30 1.52 0.25 1 0
Domestic and fancy 1958 0.741 8502 8153 0.83 2.65 0.25 1 0
glassware 1963 0.721 11943 11346 1.41 3.68 0.00 1 0
1968 0.805 15646 15646 2.13 5.45 0.00 1 0
1975 0.777 20237 21335 2.89 8.70 0.00 1 0
Upholstered furniture 1958 0.186 50918 50660 0.10 0.78 0.00 1 0
1963 0.173 61289 61786 0.12 0.93 0.00 1 0
1968 0.173 70492 70492 0.19 1.38 0.00 1 0
1975 0.276 95909 102033 0.33 2.36 0.25 1 0
Domestic furniture, other 1963 0.160 91765 92509 0.12 0.93 0.00 1 0
than upholstered 1968 0.151 119446 119446 0.19 1.38 0.00 1 0
1975 0.217 184705 193848 0.33 2.36 0.25 1 0
Upholstered divan beds, 1958 0.399 32714 31871 0.19 1.38 0.00 1 0
mattresses, and other 1963 0.418 32665 32174 0.17 1.37 0.00 1 0
bedding 1968 0.451 38611 38611 0.15 1.32 0.55 1 0
1975 0.515 44476 41789 0.18 1.52 0.55 1 0
Published books 1958 0.311 62287 58950 0.32 2.48 0.00 1 0
1963 0.339 80119 78913 0.38 2.86 0.00 1 0
1968 0.322 106922 106922 0.49 3.85 0.00 1 0

439
1975 0.328 143567 134590 0.61 5.21 0.00 1 0
Table B2 (continued)

440
Industry Year C5 SS DS K=N K=L TVCLASS ADTYPE CHANGE
Tyres and tubes, other than 1958 0.932 112412 94782 1.27 3.80 0.25 1 1
retreaded tyres 1963 0.945 144054 130074 1.54 4.06 0.25 1 1
1968 0.928 184836 184836 2.39 5.90 0.55 1 1
1975 0.964 190550 199087 2.68 7.51 0.55 1 1
Brushes and brooms 1958 0.301 17362 19193 0.19 1.43 0.25 1 0
1963 0.380 18904 17959 0.26 1.78 0.55 1 0
1968 0.425 18763 18763 0.35 2.45 0.25 1 0
1975 0.455 18479 17809 0.42 3.39 0.85 1 0
Toys and indoor games 1958 0.469 37626 36656 0.19 1.17 0.25 2 0
1963 0.477 42598 41860 0.27 1.57 0.25 2 0
1968 0.582 65106 65106 0.44 2.10 0.55 2 0
1975 0.453 93046 85285 0.54 2.69 0.55 2 0
Notes: C5 is the five-firm sales concentration ratio.
SS is sales revenue (in £1000) deflated by the general producer price index.
DS is sales revenue (in £1000) deflated by industry-specific producer price indices.
K is the value of capital stock of the corresponding MLH industry in 1980 prices (in £ million).
L is employment of the corresponding MLH industry (in 1000).
N is the number of plants with at least 25 employees of the corresponding MLH industry.
TVCLASS is a measure of the importance of TV advertising in total media advertising (see text).
ADTYPE takes the value 1 for an industry with typical or average advertising-sales ratio between 1% and 2%, and the value 2 for an industry

Appendix B
with typical or average advertising-sales ratio higher than 2%.
CHANGE takes the value 1 for an industry with a change in competition regime and 0 otherwise.
Table B3

Data Sets
Data set for the advertising regressions of chapter 5
Industry Year ADS SSDOM K=L TVTOT TVINT ADTYPE CHANGE
Cigars 1954 0.062 1659 2.36 0.00 0.00 2 0
1958 0.108 2353 3.13 0.37 0.14 2 0
1963 0.228 4077 4.84 0.43 0.18 2 0
Cigarettes 1958 0.027 172726 3.13 0.39 0.14 2 0
1963 0.051 205131 4.84 0.45 0.18 2 0
Tobacco 1958 0.039 21991 3.13 0.24 0.14 2 0
1963 0.054 25720 4.84 0.38 0.18 2 0
Spirits, other than whisky and gin 1954 0.077 6591 2.64 0.00 0.00 2 0
1958 0.088 9233 2.80 0.15 0.06 2 0
1963 0.098 13271 4.45 0.13 0.08 2 0
1968 0.080 16827 6.24 0.19 0.09 2 0
1973 0.057 38074 9.42 0.26 0.09 2 0
Wines 1954 0.077 16507 4.05 0.00 0.00 2 0
1958 0.096 19474 3.46 0.28 0.14 2 0
1963 0.106 31933 4.86 0.23 0.18 2 0
1968 0.097 41033 5.68 0.43 0.20 2 0
1973 0.045 96514 7.02 0.49 0.20 2 0
Cider and perry 1954 0.052 5038 4.05 0.00 0.00 2 0
1958 0.119 10156 3.46 0.45 0.14 2 0
1963 0.121 6975 4.86 0.63 0.18 2 0
1968 0.113 10258 5.68 0.65 0.20 2 0

441
1973 0.057 15741 7.02 0.66 0.20 2 0
Table B3 (continued)

442
Industry Year ADS SSDOM K=L TVTOT TVINT ADTYPE CHANGE
Soft drinks 1954 0.017 54884 2.71 0.00 0.00 2 0
1958 0.028 76984 2.91 0.52 0.21 2 0
1963 0.025 84746 3.04 0.47 0.27 2 0
1968 0.017 109050 3.89 0.69 0.31 2 0
1973 0.017 188960 4.79 0.74 0.31 2 0
Margarine 1954 0.037 41207 5.65 0.00 0.00 2 0
1958 0.060 37644 7.58 0.42 0.21 2 0
1963 0.088 31576 9.57 0.75 0.27 2 0
1968 0.113 22486 10.59 0.84 0.31 2 0
1973 0.083 35703 14.72 0.88 0.31 2 0
Biscuits 1954 0.007 102953 1.75 0.00 0.00 2 1
1958 0.021 105596 2.63 0.47 0.21 2 1
1963 0.021 108920 3.69 0.58 0.27 2 1
1968 0.017 110530 4.30 0.91 0.31 2 1
1973 0.021 123231 5.27 0.98 0.31 2 1
Cereal foods 1963 0.106 31031 6.45 0.80 0.27 2 0
1968 0.104 35810 9.50 0.84 0.31 2 0
1973 0.084 42952 12.10 0.85 0.31 2 0
Canned meat and poultry 1954 0.001 75672 1.33 0.00 0.00 1 0

Appendix B
1958 0.008 73115 1.43 0.44 0.21 1 0
1963 0.017 75367 1.70 0.68 0.27 1 0
1968 0.013 97236 2.44 0.80 0.31 1 0
Canned fish 1954 0.008 17726 1.33 0.00 0.00 1 0

Data Sets
1958 0.011 40190 1.43 0.36 0.21 1 0
1963 0.012 30510 1.70 0.22 0.27 1 0
1968 0.006 41185 2.44 0.89 0.31 1 0
Meat and fish pastes 1954 0.039 5240 1.33 0.00 0.00 2 0
1958 0.041 6913 1.43 0.52 0.21 2 0
1963 0.015 7432 1.70 0.78 0.27 2 0
Canned vegetables 1954 0.013 51854 1.63 0.00 0.00 1 0
1958 0.015 61018 1.95 0.55 0.21 1 0
1963 0.017 68976 2.99 0.44 0.27 1 0
1968 0.017 78895 4.28 0.86 0.31 1 0
Frozen foods 1954 0.010 8693 1.48 0.00 0.00 2 0
1958 0.030 18688 1.67 0.71 0.21 2 0
1963 0.030 58854 2.26 0.74 0.27 2 0
1968 0.025 91278 3.23 0.83 0.31 2 0
Ice cream and ice lollies 1954 0.017 14817 3.72 0.00 0.00 2 0
1958 0.028 19234 3.65 0.63 0.14 2 0
1963 0.051 13726 4.74 0.77 0.18 2 0
1968 0.057 16717 5.81 0.72 0.20 2 0
1973 0.014 44027 7.34 0.57 0.20 2 0
Canned milk 1954 0.032 12178 3.72 0.00 0.00 2 1
1958 0.038 16747 3.65 0.35 0.21 2 1
1963 0.039 18249 4.74 0.35 0.27 2 1
1968 0.026 18820 5.81 0.81 0.31 2 1
1973 0.014 20198 7.34 0.99 0.31 2 1

443
Table B3 (continued)

444
Industry Year ADS SSDOM K=L TVTOT TVINT ADTYPE CHANGE
Potato crisps 1963 0.028 19905 2.99 0.94 0.27 2 0
1968 0.024 26886 4.28 0.87 0.31 2 0
Table jellies 1954 0.034 7042 1.63 0.00 0.00 2 0
1958 0.036 7170 1.95 0.83 0.06 2 0
1963 0.015 6411 2.99 0.62 0.08 2 0
1968 0.013 7101 4.28 0.12 0.09 2 0
Jams and marmalade 1954 0.010 38432 1.63 0.00 0.00 1 0
1958 0.010 39126 1.95 0.45 0.14 1 0
1963 0.027 38412 2.99 0.51 0.18 1 0
1968 0.016 34622 4.28 0.63 0.20 1 0
Sauces, pickles, and salad creams 1954 0.022 19178 1.63 0.00 0.00 2 0
1958 0.027 20709 1.95 0.40 0.14 2 0
1963 0.023 24460 2.99 0.21 0.18 2 0
1968 0.027 26367 4.28 0.42 0.20 2 0
Soups 1954 0.051 11652 1.63 0.00 0.00 2 0
1958 0.043 23580 1.95 0.49 0.21 2 0
1963 0.057 33115 2.99 0.45 0.27 2 0
1968 0.048 33233 4.28 0.76 0.31 2 0
Coffee and coffee extracts 1954 0.044 12625 2.88 0.00 0.00 2 0

Appendix B
1958 0.074 19922 3.82 0.48 0.21 2 0
1963 0.084 27602 5.27 0.63 0.27 2 0
1968 0.024 40452 8.07 0.74 0.31 2 0
1973 0.040 53582 9.64 0.84 0.31 2 0
Cocoa and drinking chocolate 1954 0.071 5796 1.95 0.00 0.00 2 1

Data Sets
1958 0.116 4625 2.42 0.40 0.21 2 1
1963 0.081 4442 3.40 0.63 0.27 2 1
1968 0.057 4415 4.64 0.90 0.31 2 1
1973 0.059 3652 5.90 1.00 0.31 2 1
Chocolate confectionery 1954 0.014 131442 1.95 0.00 0.00 2 1
1958 0.024 122200 2.42 0.57 0.21 2 1
1963 0.059 123018 3.40 0.77 0.27 2 1
1968 0.060 134432 4.64 0.98 0.31 2 1
1973 0.045 160991 5.90 0.95 0.31 2 1
Adhesive dressings 1954 0.081 2309 1.42 0.00 0.00 2 1
1958 0.077 2276 2.01 0.31 0.21 2 1
1963 0.069 3906 3.92 0.36 0.27 2 1
1968 0.039 5330 6.07 0.86 0.31 2 1
1973 0.021 4879 6.26 0.97 0.31 2 1
Sanitary towels 1954 0.019 8015 1.42 0.00 0.00 2 0
1958 0.013 9331 2.01 0.00 0.00 2 0
1963 0.027 10474 3.92 0.00 0.00 2 0
1968 0.031 11941 6.07 0.00 0.00 2 0
1973 0.060 11898 6.26 0.00 0.00 2 0
Dental preparations 1954 0.241 6445 1.98 0.00 0.00 2 0
1958 0.224 8703 2.02 0.54 0.21 2 0
1963 0.274 11123 2.03 0.87 0.27 2 0
1968 0.167 11713 2.96 0.81 0.31 2 0
1973 0.165 17081 4.27 0.85 0.31 2 0

445
Table B3 (continued)

446
Industry Year ADS SSDOM K=L TVTOT TVINT ADTYPE CHANGE
Hair preparations 1954 0.287 10239 1.98 0.00 0.00 2 0
1958 0.324 13267 2.02 0.50 0.14 2 0
1963 0.258 22605 2.03 0.40 0.18 2 0
1968 0.179 26631 2.96 0.47 0.20 2 0
1973 0.139 42914 4.27 0.63 0.20 2 0
Toilet soap 1954 0.105 12406 6.78 0.00 0.00 2 0
1958 0.155 15296 7.35 0.39 0.21 2 0
1963 0.190 15777 9.14 0.87 0.27 2 0
1968 0.139 15878 14.43 0.93 0.31 2 0
1973 0.066 24723 15.69 0.87 0.31 2 0
Other toilet preparations 1954 0.140 21117 1.98 0.00 0.00 2 0
1958 0.145 27282 2.02 0.28 0.06 2 0
1963 0.114 44477 2.03 0.26 0.08 2 0
1968 0.110 62266 2.96 0.27 0.09 2 0
1973 0.091 96613 4.27 0.51 0.09 2 0
Razors and blades 1954 0.071 5294 1.47 0.00 0.00 2 0
1958 0.087 6007 1.75 0.51 0.21 2 0
1963 0.115 6923 2.20 0.53 0.27 2 0
1968 0.222 7024 3.42 0.74 0.31 2 0

Appendix B
1973 0.117 7462 4.19 0.75 0.31 2 0
Toilet brushes 1954 0.060 2946 1.21 0.00 0.00 2 0

Data Sets
1958 0.090 2458 1.43 0.33 0.14 2 0
1963 0.039 3291 1.78 0.44 0.18 2 0
1968 0.027 3262 2.45 0.55 0.20 2 0
1973 0.020 4258 2.99 0.68 0.20 2 0
Paper handkerchiefs and facial tissues 1963 0.080 5449 4.84 0.45 0.18 2 0
1968 0.044 7833 6.08 0.84 0.20 2 0
1973 0.016 10665 7.90 0.43 0.20 2 0
Paper towels 1963 0.042 2670 4.84 0.03 0.18 2 0
1968 0.038 5455 6.08 0.66 0.20 2 0
1973 0.023 9012 7.90 0.85 0.20 2 0
Toilet paper 1963 0.039 12292 4.84 0.46 0.18 2 0
1968 0.009 17165 6.08 0.19 0.20 2 0
1973 0.008 28020 7.90 0.68 0.20 2 0
Domestic brushes and paintbrushes 1954 0.008 11114 1.21 0.00 0.00 1 0
1958 0.016 12261 1.43 0.27 0.06 1 0
1963 0.016 13675 1.78 0.49 0.08 1 0
1968 0.012 13171 2.45 0.10 0.09 1 0
1973 0.009 14023 2.99 0.28 0.09 1 0
Paint 1954 0.008 115533 2.64 0.00 0.00 2 1
1958 0.019 127330 3.19 0.24 0.14 2 1
1963 0.026 125338 3.64 0.39 0.18 2 1
1968 0.021 119802 4.37 0.56 0.20 2 1
1973 0.021 165041 5.52 0.63 0.20 2 1

447
Table B3 (continued)

448
Industry Year ADS SSDOM K=L TVTOT TVINT ADTYPE CHANGE
Soap and detergents 1954 0.083 59746 6.78 0.00 0.00 2 0
1958 0.138 65331 7.35 0.56 0.21 2 0
1963 0.132 67721 9.14 0.89 0.27 2 0
1968 0.095 73928 14.43 0.98 0.31 2 0
1973 0.067 92137 15.69 0.99 0.31 2 0
Cleansers 1954 0.268 3561 6.78 0.00 0.00 2 0
1958 0.337 6919 7.35 0.49 0.21 2 0
1963 0.337 9979 9.14 0.88 0.27 2 0
1968 0.333 8940 14.43 0.87 0.31 2 0
1973 0.167 11041 15.69 0.94 0.31 2 0
Floor and furniture polishes 1954 0.068 6596 1.69 0.00 0.00 2 0
1958 0.105 7478 2.20 0.45 0.21 2 0
1963 0.132 6927 3.04 0.43 0.27 2 0
1968 0.123 7791 4.29 0.75 0.31 2 0
1973 0.104 8024 5.95 0.96 0.31 2 0
Shoe polishes 1954 0.088 2508 1.69 0.00 0.00 2 0
1958 0.080 2864 2.20 0.23 0.06 2 0
1963 0.140 3186 3.04 0.23 0.08 2 0
1968 0.066 2909 4.29 0.57 0.09 2 0

Appendix B
1973 0.035 2512 5.95 0.01 0.09 2 0
Disinfectants and household deodorisers 1954 0.080 5768 2.54 0.00 0.00 2 0

Data Sets
1958 0.071 8564 2.96 0.54 0.21 2 0
1963 0.051 8194 4.56 0.58 0.27 2 0
1968 0.045 10840 7.49 0.80 0.31 2 0
1973 0.019 17517 14.92 0.81 0.31 2 0
Insecticides and pesticides 1954 0.024 7879 2.54 0.00 0.00 2 0
1958 0.026 11365 2.96 0.26 0.06 2 0
1963 0.034 11360 4.56 0.16 0.08 2 0
1968 0.029 12445 7.49 0.24 0.09 2 0
Dog and cat foods 1954 0.051 11133 2.38 0.00 0.00 2 0
1958 0.068 22702 2.77 0.53 0.21 2 0
1963 0.101 35482 3.88 0.86 0.27 2 0
1968 0.101 52577 6.16 0.98 0.31 2 0
1973 0.042 70972 8.81 0.95 0.31 2 0
Lawn mowers 1954 0.034 3030 1.87 0.00 0.00 2 0
1958 0.034 4448 2.07 0.03 0.00 2 0
1963 0.025 6908 2.23 0.00 0.00 2 0
1968 0.024 13963 2.82 0.03 0.00 2 0
1973 0.017 22309 3.65 0.04 0.00 2 0
Rubber gloves 1963 0.038 2469 0.73 0.00 0.27 2 0
1968 0.025 3337 0.92 0.98 0.31 2 0
1973 0.011 4424 1.20 0.99 0.31 2 0
Sewing machines 1958 0.137 2750 2.07 0.00 0.06 2 0
1963 0.117 2913 2.23 0.04 0.08 2 0
1968 0.094 3630 2.82 0.32 0.09 2 0

449
Table B3 (continued)

450
Industry Year ADS SSDOM K=L TVTOT TVINT ADTYPE CHANGE
Notepaper 1958 0.053 3122 1.76 0.02 0.00 2 1
1963 0.028 3907 2.58 0.19 0.00 2 1
1968 0.027 3050 3.36 0.00 0.00 2 1
Pens and mechanical pencils 1954 0.076 6633 1.77 0.00 0.00 2 0
1958 0.068 7969 1.63 0.29 0.06 2 0
1963 0.047 7925 2.33 0.00 0.08 2 0
1968 0.047 11423 3.41 0.18 0.09 2 0
1973 0.037 13506 5.18 0.26 0.09 2 0
Lead pencils 1954 0.009 1495 1.77 0.00 0.00 1 1
1958 0.018 1787 1.63 0.00 0.00 1 1
1963 0.014 1764 2.33 0.00 0.00 1 1
1968 0.006 1402 3.41 0.00 0.00 1 1
Toys and games 1963 0.015 40090 1.57 0.34 0.18 1 0
1968 0.020 59518 2.10 0.55 0.20 1 0
1973 0.016 95030 2.72 0.67 0.20 1 0
Baby carriages 1954 0.008 5616 0.94 0.00 0.00 1 0
1958 0.017 6052 1.17 0.47 0.00 1 0
1963 0.011 6965 1.57 0.11 0.00 1 0
1968 0.003 5963 2.10 0.00 0.00 1 0

Appendix B
1973 0.001 5771 2.72 0.00 0.00 1 0
Typewriters 1954 0.031 3978 2.83 0.00 0.00 2 1

Data Sets
1958 0.034 5091 2.64 0.03 0.00 2 1
1963 0.019 7315 3.14 0.01 0.00 2 1
1968 0.009 10096 4.48 0.00 0.00 2 1
Tape recorders 1958 0.038 3531 1.10 0.03 0.00 2 0
1963 0.024 9505 1.00 0.00 0.00 2 0
1968 0.023 6023 1.98 0.00 0.00 2 0
Radio and TV sets 1954 0.020 73585 0.97 0.00 0.00 1 0
1958 0.024 82120 1.10 0.04 0.00 1 0
1963 0.019 84998 1.00 0.04 0.00 1 0
1968 0.011 95841 1.98 0.05 0.00 1 0
1973 0.006 307292 2.32 0.13 0.00 1 0
Office furniture 1963 0.016 20428 1.87 0.00 0.00 1 0
1968 0.010 26308 2.24 0.00 0.00 1 0
Electric lamps and fluorescent lighting 1954 0.011 18503 1.64 0.00 0.00 1 1
1958 0.017 19963 1.72 0.29 0.06 1 1
1963 0.013 30653 2.03 0.12 0.08 1 1
1968 0.015 35515 2.68 0.32 0.09 1 1
1973 0.008 35630 3.71 0.04 0.09 1 1
Mattresses, divan beds, pillows 1954 0.008 45563 1.31 0.00 0.00 1 0
1958 0.008 45168 1.38 0.08 0.14 1 0
1963 0.010 37459 1.37 0.02 0.18 1 0
1968 0.011 47437 1.32 0.50 0.20 1 0
1973 0.010 59478 1.39 0.46 0.20 1 0

451
Table B3 (continued)

452
Industry Year ADS SSDOM K=L TVTOT TVINT ADTYPE CHANGE
Car batteries and accumulators 1954 0.016 22165 1.64 0.00 0.00 1 1
1958 0.012 21927 1.72 0.15 0.06 1 1
1963 0.011 26945 2.03 0.15 0.08 1 1
1968 0.009 34531 2.68 0.32 0.09 1 1
1973 0.002 45027 3.71 0.40 0.09 1 1
Tyres and tubes 1954 0.017 68407 2.88 0.00 0.00 1 1
1958 0.021 72141 3.80 0.17 0.14 1 1
1963 0.017 97444 4.06 0.22 0.18 1 1
1968 0.016 146660 5.90 0.40 0.20 1 1
1973 0.012 148535 7.87 0.47 0.20 1 1
Motorcycles, scooters, and mopeds 1954 0.012 12564 4.57 0.00 0.00 1 1
1958 0.023 16998 5.13 0.13 0.00 1 1
1963 0.023 10312 7.06 0.00 0.00 1 1
1968 0.014 7847 8.08 0.00 0.00 1 1
1973 0.015 15697 7.74 0.00 0.00 1 1
Cars 1954 0.011 180093 6.68 0.00 0.00 1 0
1958 0.012 255515 6.46 0.01 0.00 1 0
1963 0.012 415130 5.81 0.01 0.00 1 0
1968 0.010 537892 6.49 0.03 0.00 1 0

Appendix B
1973 0.012 906758 6.44 0.09 0.00 1 0
Industrial trucks and tractors 1963 0.021 10006 1.56 0.00 0.00 1 0
1968 0.014 30320 1.91 0.00 0.00 1 0
Men’s shirts, collars, and pyjamas 1954 0.006 46871 0.29 0.00 0.00 1 0

Data Sets
1958 0.014 48898 0.34 0.41 0.06 1 0
1963 0.014 54609 0.47 0.18 0.08 1 0
1968 0.012 55695 0.67 0.19 0.09 1 0
1973 0.004 69034 0.87 0.00 0.09 1 0
Women’s hosiery 1954 0.026 20947 3.76 0.00 0.00 1 0
1958 0.017 47937 4.26 0.09 0.06 1 0
1963 0.011 53552 4.23 0.09 0.08 1 0
1968 0.010 58079 4.92 0.18 0.09 1 0
1973 0.008 41551 5.91 0.30 0.09 1 0
Corsetry 1954 0.034 21044 0.88 0.00 0.00 2 0
1958 0.062 27406 0.85 0.18 0.14 2 0
1963 0.047 35272 0.91 0.03 0.18 2 0
1968 0.048 35927 1.06 0.46 0.20 2 0
1973 0.033 36739 1.21 0.84 0.20 2 0
Footwear 1954 0.009 199319 0.58 0.00 0.00 1 0
1958 0.010 184301 0.67 0.14 0.06 1 0
1963 0.010 214995 0.87 0.14 0.08 1 0
1968 0.011 214349 1.15 0.40 0.09 1 0
1973 0.006 269970 1.47 0.20 0.09 1 0
Domestic refrigerators and deep freezers 1958 0.037 14120 2.55 0.20 0.00 2 1
1963 0.032 16587 2.27 0.27 0.00 2 1
1968 0.011 22095 3.00 0.01 0.00 2 1
1973 0.022 31393 4.06 0.16 0.00 2 1

453
Table B3 (continued)

454
Industry Year ADS SSDOM K=L TVTOT TVINT ADTYPE CHANGE
Washing machines 1954 0.025 19213 2.19 0.00 0.00 2 1
1958 0.044 26214 2.55 0.22 0.06 2 1
1968 0.029 29250 3.00 0.33 0.09 2 1
1973 0.008 52081 4.06 0.26 0.09 2 1
Electric cookers and heaters 1954 0.016 18198 2.19 0.00 0.00 2 1
1958 0.033 24089 2.55 0.08 0.00 2 1
1963 0.042 44342 2.27 0.14 0.00 2 1
1968 0.018 52265 3.00 0.05 0.00 2 1
1973 0.009 61866 4.06 0.14 0.00 2 1

Notes: ADS is the advertising-sales ratio.


SSDOM is sales revenue in the UK market (in £1000) deflated by the general producer price index.
K is the value of capital stock of the corresponding MLH industry in 1980 prices (in £ million).
L is employment of the corresponding MLH industry (in 1000).
TVTOT is the fraction of TV advertising in total media advertising.
TVINT is an alternative measure of the significance of TV advertising (see text).
ADTYPE takes the value 1 for an industry with average advertising-sales ratio between 1% and 2%, and 2 for an industry with average
advertising-sales ratio higher than 2% over 1958–1968.
CHANGE takes the value 1 for an industry with a change in competition regime and 0 otherwise.

Appendix B
Table B4

Data Sets
Data set for the innovation regressions of chapter 6
Industry Period INN1 INN2 INN3 SS DS K=N K=L RDTYPE CHANGE
Lubricating oils and greases (MLH 1952–56 3 4 3 58551 63111 0.49 4.03 1 0
263) 1957–61 2 1 2 71533 74920 0.70 4.88 1 0
1962–66 0 0 0 74886 74886 0.80 6.01 1 0
1967–71 2 2 2 80197 83787 1.27 10.86 1 0
1972–76 0 0 0 103190 108339 1.60 11.97 1 0
General chemicals, organic, other 1952–56 0 0 0 98593 81489 5.00 21.25 2 0
than coal tar products (part of 1957–61 4 5 0 166999 143385 8.25 27.23 2 0
MLH 271.2)
1962–66 4 3 3 240323 240323 9.66 25.75 2 0
1967–71 3 4 3 312446 354349 16.29 49.23 2 0
1972–76 3 2 1 420695 551838 25.05 68.45 2 0
General chemicals, other than 1952–56 5 5 1 85966 71052 2.40 10.10 1 0
inorganic and organic (MLH 271.3) 1957–61 4 3 2 143114 122877 3.99 12.84 1 0
1962–66 4 4 2 168424 168424 4.67 17.77 1 0
1967–71 3 3 2 215759 244695 6.47 25.04 1 0
1972–76 5 9 2 280777 363533 7.59 26.70 1 0
Pharmaceutical chemicals and 1952–56 7 6 4 119569 81112 0.84 3.40 2 0
preparations (MLH 272) 1957–61 19 22 7 141742 120695 1.32 4.10 2 0
1962–66 15 9 7 194889 194889 1.68 4.93 2 0
1967–71 12 14 7 244569 300118 2.67 7.55 2 0
1972–76 12 16 8 363509 525543 4.46 11.30 2 0

455
Table B4 (continued)

456
Industry Period INN1 INN2 INN3 SS DS K=N K=L RDTYPE CHANGE
Paint (MLH 274) 1952–56 4 4 3 116353 109907 0.40 2.69 1 1
1957–61 6 8 2 132779 128358 0.57 3.19 1 1
1962–66 2 1 1 136354 136354 0.71 3.64 1 1
1967–71 4 4 4 128602 144445 0.92 4.37 1 1
1972–76 4 3 4 171899 193160 1.19 5.52 1 1
Soap and detergents (MLH 275) 1952–56 3 1 2 98982 104347 1.74 6.64 1 0
1957–61 3 3 0 111817 108659 2.27 7.41 1 0
1962–66 4 4 2 118241 118241 3.53 9.14 1 0
1967–71 3 3 1 128165 137219 4.44 14.43 1 0
1972–76 1 1 1 156682 173380 3.81 15.69 1 0
Synthetic resins and plastics 1952–56 9 8 1 109462 68260 4.66 13.02 2 0
materials and synthetic rubber 1957–61 11 14 3 171822 135849 5.12 12.18 2 0
(MLH 276) 1962–66 14 12 4 258679 258679 6.90 16.68 2 0
1967–71 8 8 5 366558 446994 7.90 19.17 2 0
1972–76 8 7 5 399964 610834 10.76 31.98 2 0
Dyestuffs and pigments (MLH 277) 1952–56 6 5 4 80988 77745 3.56 8.44 2 0
1957–61 11 11 5 72350 68237 4.79 9.73 2 0
1962–66 3 3 2 82818 82818 5.79 12.00 2 0
1967–71 11 11 7 89178 90339 10.71 20.56 2 0

Appendix B
1972–76 6 8 4 139149 157244 13.75 26.88 2 0
Formulated adhesives, gelatine, 1952–56 1 1 0 18435 16138 0.52 4.18 2 0

Data Sets
glue, size, etc. (MLH 279.2) 1957–61 0 0 0 19423 18319 0.70 5.57 2 0
1962–66 2 2 1 25493 25493 0.96 7.22 2 0
1967–71 3 3 2 32410 31232 1.39 7.99 2 0
1972–76 3 3 2 37579 36640 1.98 11.33 2 0
Formulated pesticides and 1952–56 0 0 0 20028 16106 0.37 2.27 2 0
disinfectants (MLH 279.4) 1957–61 2 3 2 25051 21331 0.40 2.72 2 0
1962–66 2 1 1 30600 30600 0.68 4.56 2 0
1967–71 2 2 2 38867 42791 1.28 7.49 2 0
1972–76 6 9 6 52194 72446 2.69 14.92 2 0
Surgical bandages, etc., and 1952–56 0 0 0 25749 19997 0.28 1.31 0 1
sanitary towels (MLH 279.6) 1957–61 1 2 1 23513 20804 0.55 2.01 0 1
1962–66 4 3 1 31660 31660 1.16 3.92 0 1
1967–71 2 2 0 37314 36288 2.12 6.07 0 1
1972–76 1 1 0 45700 42215 2.47 6.26 0 1
Aluminium and aluminium alloys 1952–56 8 6 8 143074 160803 1.05 3.23 0 1
(MLH 321) 1957–61 11 11 10 152853 158305 1.39 4.27 0 1
1962–66 7 6 6 192061 192061 1.55 5.02 0 1
1967–71 7 7 4 247753 232792 2.22 7.32 0 1
1972–76 5 4 5 279127 266303 4.66 15.66 0 1
Copper, brass and other copper 1952–56 0 0 0 301232 225263 0.92 4.44 0 1
alloys (MLH 322) 1957–61 0 1 0 241120 265403 1.26 5.22 0 1
1962–66 2 1 0 301753 301753 1.35 5.80 0 1
1967–71 2 3 2 420605 243113 1.64 6.71 0 1
1972–76 1 0 1 417250 220994 1.87 7.31 0 1

457
Table B4 (continued)

458
Industry Period INN1 INN2 INN3 SS DS K=N K=L RDTYPE CHANGE
Miscellaneous nonferrous metals 1952–56 4 4 4 140387 134577 1.56 6.21 0 0
(MLH 323) 1957–61 2 2 1 143488 150113 1.74 6.70 0 0
1962–66 2 2 1 174836 174836 1.83 7.17 0 0
1967–71 0 1 0 233154 178420 2.12 9.81 0 0
1972–76 2 2 0 207510 155632 2.63 14.65 0 0
Welding and flame-cutting 1952–56 1 2 1 12134 13683 0.65 3.06 1 0
equipment (part of MLH 332) 1957–61 5 4 4 18226 18627 0.80 3.54 1 0
1962–66 6 6 3 23137 23137 0.77 3.26 1 0
1967–71 2 2 0 30380 30802 0.96 4.02 1 0
1972–76 4 5 2 31123 29274 1.02 5.08 1 0
Industrial engines (MLH 334) 1952–56 0 0 0 62369 58111 2.03 3.86 1 1
1957–61 1 1 1 67277 56870 2.79 4.07 1 1
1962–66 0 0 0 85543 85543 3.02 4.11 1 1
1967–71 0 1 0 101817 96445 3.53 4.74 1 1
1972–76 3 3 3 104037 92112 5.10 6.08 1 1
Powered industrial trucks and 1952–56 0 0 0 8102 10043 0.22 1.13 1 0
industrial tractors (part of MLH 1957–61 0 0 0 11145 12673 0.29 1.41 1 0
337)
1962–66 2 2 2 17838 17838 0.33 1.56 1 0
1967–71 1 2 1 46045 49343 0.41 1.91 1 0

Appendix B
1972–76 2 1 1 70049 70136 0.57 2.44 1 0
Mining machinery (MLH 339.1) 1952–56 11 8 7 36875 41582 0.49 2.10 0 0

Data Sets
1957–61 10 14 8 53516 54693 0.50 2.09 0 0
1962–66 28 28 21 63789 63789 0.56 2.38 0 0
1967–71 24 25 18 71840 72838 0.70 2.81 0 0
1972–76 17 13 13 86122 82165 0.98 3.50 0 0
Printing, bookbinding, and paper 1952–56 3 3 3 29191 32917 0.30 1.91 1 0
goods making machinery (MLH 1957–61 2 3 2 25392 25951 0.38 2.15 1 0
339.2)
1962–66 8 14 5 35795 35795 0.49 2.43 1 0
1967–71 22 18 16 44586 45206 0.67 2.98 1 0
1972–76 13 10 7 50969 49067 0.92 4.10 1 0
Space-heating, ventilating, and air- 1952–56 4 6 4 31097 35067 0.26 1.76 0 0
conditioning equipment (MLH 1957–61 7 8 4 44873 45860 0.37 2.13 0 0
339.4) 1962–66 12 10 9 77232 77232 0.36 2.06 0 0
1967–71 4 3 1 95937 97270 0.41 2.40 0 0
1972–76 5 5 4 125158 120486 0.54 3.11 0 0
Packaging and bottling machinery 1952–56 1 1 1 7929 8942 0.30 1.85 0 0
(MLH 339.8) 1957–61 1 1 1 16473 16835 0.42 2.07 0 0
1962–66 2 5 2 19260 19260 0.49 2.23 0 0
1967–71 6 5 4 24115 24450 0.59 2.82 0 0
1972–76 4 4 4 30092 29197 0.65 3.65 0 0
Miscellaneous machinery, other 1952–56 11 13 8 93720 105683 0.30 1.85 0 0
than rolling mills (part of MLH 1957–61 17 15 12 113597 116097 0.42 2.07 0 0
339.9) 1962–66 19 21 9 138562 138562 0.49 2.23 0 0
1967–71 18 16 14 154218 156362 0.59 2.82 0 0
1972–76 12 17 11 228744 221406 0.65 3.65 0 0

459
Table B4 (continued)

460
Industry Period INN1 INN2 INN3 SS DS K=N K=L RDTYPE CHANGE
Constructional and other fabricated 1952–56 2 2 1 91924 103659 0.55 2.49 0 1
ironwork and steelwork (part of 1957–61 4 6 4 140853 143952 0.64 2.50 0 1
MLH 341) 1962–66 9 7 8 139156 139156 0.70 2.60 0 1
1967–71 1 1 1 170178 172544 0.77 2.80 0 1
1972–76 1 2 1 242387 233340 0.80 3.31 0 1
Photographic and document 1952–56 5 8 4 7762 6920 0.30 2.86 2 0
copying equipment (MLH 351) 1957–61 5 3 4 8318 7941 0.66 3.29 2 0
1962–66 3 2 2 12804 12804 0.54 3.10 2 0
1967–71 5 7 2 41956 39170 1.02 2.86 2 0
1972–76 8 6 2 57154 57442 1.51 5.64 2 0
Machinery for generating, 1952–56 7 9 6 82325 73401 0.91 1.64 2 1
transmitting and distributing 1957–61 7 5 7 99031 94553 1.12 1.87 2 1
electric power (part of MLH 361) 1962–66 4 7 4 120070 120070 1.10 1.98 2 1
1967–71 4 1 4 151267 157054 1.28 3.04 2 1
1972–76 3 3 3 145302 153358 1.35 3.76 2 1
Switchgear and controlling gear for 1952–56 1 1 1 80736 71984 0.91 1.64 2 1
motors (part of MLH 361) 1957–61 3 3 3 118969 113589 1.12 1.87 2 1
1962–66 2 3 2 139233 139233 1.10 1.98 2 1
1967–71 5 4 5 144897 150440 1.28 3.04 2 1

Appendix B
1972–76 2 2 2 115723 122139 1.35 3.76 2 1
Other electrical machinery (part of 1952–56 2 2 2 68706 61258 0.91 1.64 1 1

Data Sets
MLH 361) 1957–61 7 8 6 89776 85716 1.12 1.87 1 1
1962–66 6 4 4 89770 89770 1.10 1.98 1 1
1967–71 2 2 2 99909 103731 1.28 3.04 1 1
1972–76 1 2 1 99076 104570 1.35 3.76 1 1
Insulated wires and cables, other 1952–56 5 5 2 101215 90243 1.93 3.22 0 1
than telecommunication cables 1957–61 4 4 3 112680 107585 3.00 4.25 0 1
(part of MLH 362)
1962–66 1 1 1 148034 148034 2.84 4.68 0 1
1967–71 3 5 2 171854 152860 3.75 5.42 0 1
1972–76 6 4 6 188044 169877 5.12 7.57 0 1
Radio and electronic components 1952–56 19 23 6 74132 66096 0.44 1.62 2 0
(MLH 364) 1957–61 32 30 13 112946 107839 0.53 1.80 2 0
1962–66 19 19 7 151737 151737 0.60 1.63 2 0
1967–71 28 28 15 220987 229441 0.85 2.40 2 0
1972–76 23 30 16 353699 373310 1.36 3.35 2 0
Broadcast receiving and sound 1952–56 12 11 8 81867 72992 0.44 0.99 1 0
reproducing equipment, except 1957–61 10 11 8 101267 96688 0.53 1.10 1 0
gramophone records (part of MLH
1962–66 10 12 9 104905 104905 0.55 1.00 1 0
365)
1967–71 10 8 4 108476 112625 1.01 1.98 1 0
1972–76 2 2 1 262529 277085 1.39 2.32 1 0
Electronic computers (MLH 366) 1952–56 8 11 8 5305 4717 1.58 4.40 2 0
1957–61 8 9 4 7433 7097 1.84 4.83 2 0
1962–66 9 8 4 40141 40141 1.80 4.68 2 0
1967–71 7 6 5 100799 104655 2.30 4.21 2 0
1972–76 12 13 8 206781 218246 2.67 4.59 2 0

461
Table B4 (continued)

462
Industry Period INN1 INN2 INN3 SS DS K=N K=L RDTYPE CHANGE
Radio, radar and electronic capital 1952–56 7 9 6 50556 45075 1.30 2.80 2 0
goods (MLH 367) 1957–61 13 14 8 51668 49332 1.56 3.10 2 0
1962–66 21 20 17 127298 127298 1.37 2.80 2 0
1967–71 24 28 23 157694 163727 1.37 3.32 2 0
1972–76 28 31 26 218401 230510 1.65 3.64 2 0
Electric appliances primarily for 1952–56 1 1 1 89989 73899 0.64 2.11 1 1
domestic use (MLH 368) 1957–61 4 4 4 133477 118774 1.18 2.55 1 1
1962–66 0 1 0 200274 200274 1.18 2.27 1 1
1967–71 1 0 1 202937 227281 1.43 3.00 1 1
1972–76 1 1 0 231079 289800 2.18 4.06 1 1
Electrical equipment for motor 1952–56 2 2 1 60573 61414 0.41 1.64 1 0
vehicles, cycles, and aircraft (part 1957–61 3 3 2 71395 77665 0.49 1.68 1 0
of MLH 369) 1962–66 8 7 5 84342 84342 0.63 2.03 1 0
1967–71 3 3 2 108498 109530 0.98 2.68 1 0
1972–76 0 1 0 130750 134105 1.56 3.71 1 0
Primary batteries and accumulators 1952–56 0 0 0 38407 37471 0.41 1.64 1 1
(part of MLH 369) 1957–61 1 1 0 39663 41494 0.49 1.68 1 1
1962–66 0 0 0 49182 49182 0.63 2.03 1 1
1967–71 1 1 0 60498 56122 0.98 2.68 1 1

Appendix B
1972–76 1 1 1 71920 64716 1.56 3.71 1 1
Electric lamps, electric light 1952–56 2 1 2 58326 51798 0.41 1.64 1 1

Data Sets
fittings, and wiring accessories 1957–61 0 1 0 70465 64431 0.49 1.68 1 1
(part of MLH 369) 1962–66 5 5 3 97992 97992 0.63 2.03 1 1
1967–71 6 6 6 110277 116044 0.98 2.68 1 1
1972–76 3 4 3 136861 148468 1.56 3.71 1 1
Marine machinery (part of MLH 1952–56 5 3 4 109362 123322 0.43 0.98 1 0
370) 1957–61 3 5 1 78298 80021 0.59 1.25 1 0
1962–66 7 5 6 65119 65119 0.86 1.96 1 0
1967–71 2 3 2 71559 72553 0.98 2.11 1 0
1972–76 1 0 1 78174 76500 1.36 2.68 1 0
Motor vehicles and parts, except 1952–56 12 11 9 824394 899139 2.72 6.72 1 0
trailers and caravans (part of MLH 1957–61 16 18 13 1172390 1186438 3.47 6.46 1 0
381) 1962–66 14 14 8 1612064 1612064 3.49 5.81 1 0
1967–71 19 23 16 1820051 1869719 3.71 6.49 1 0
1972–76 21 23 19 2049130 2012628 4.01 6.43 1 0
Motorcycles (part of MLH 382) 1952–56 1 0 1 35061 31387 1.50 4.44 1 1
1957–61 1 1 1 27233 21586 1.81 5.13 1 1
1962–66 0 0 0 16224 16224 2.84 7.06 1 1
1967–71 1 1 1 18871 17139 3.65 8.08 1 1
1972–76 2 3 0 16263 14118 3.34 7.74 1 1
Paper and board (MLH 481) 1952–56 6 6 4 293269 269608 1.66 5.83 0 1
1957–61 5 8 5 321880 296653 2.74 8.74 0 1
1962–66 11 8 10 350074 350074 3.23 9.61 0 1
1967–71 5 6 3 377973 390695 4.91 13.74 0 1
1972–76 3 3 3 396899 408918 6.13 17.72 0 1

463
Table B4 (continued)

464
Industry Period INN1 INN2 INN3 SS DS K=N K=L RDTYPE CHANGE
Tyres and tubes (part of MLH 491) 1952–56 0 0 0 93731 99189 0.92 2.69 1 1
1957–61 1 1 0 96312 89924 1.26 3.80 1 1
1962–66 0 0 0 123440 123440 1.54 4.06 1 1
1967–71 1 3 1 158386 175366 2.39 5.90 1 1
1972–76 3 1 3 162691 182942 3.03 7.87 1 1
Plastic products (MLH 496) 1952–56 7 8 5 41734 35482 0.23 1.66 0 0
1957–61 6 6 4 65580 60985 0.26 1.86 0 0
1962–66 9 9 7 146583 146583 0.33 2.24 0 0
1967–71 18 24 15 249666 321945 0.55 3.60 0 0
1972–76 17 11 13 397786 574061 0.82 5.26 0 0
Notes: The classification according to MLH industries follows the 1968 Standard Industrial Classification.
INN1 is the number of innovations for time periods 1952–56, 1957–61, 1962–66, 1967–71, and 1972–76.
INN2 is the number of innovations for time periods 1953–57, 1958–62, 1963–67, 1968–72, and 1973–77.
INN3 is the number of innovations, excluding those reported as originating outside the UK, for time periods 1952–56, 1957–61, 1962–66, 1967–
71, and 1972–76.
SS is sales revenue (in £1000) deflated by the general producer price index for the years 1953, 1958, 1963, 1968, and 1973.
DS is sales revenue (in £1000) deflated by industry-specific producer price indices for the years 1953, 1958, 1963, 1968, and 1973.
K is the value of capital stock in 1980 prices (in £ million) of the corresponding MLH industry for the years 1953, 1958, 1963, 1968, and 1973.
L is employment (in 1000) of the corresponding MLH industry for the years 1953, 1958, 1963, 1968, and 1973.
N is the number of plants with at least 25 employees of the corresponding MLH industry for the years 1953, 1958, 1963, 1968, and 1973.
RDTYPE takes the value 0 for an industry with typical R&D-sales ratio lower than 1%, 1 for an industry with typical R&D-sales ratio between

Appendix B
1% and 2%, and 2 for an industry with typical R&D-sales ratio higher than 2%.
CHANGE takes the value 1 for an industry with a change in competition regime and 0 otherwise.
Table B5

Data Sets
Data set for the concentration regressions of chapter 6
Industry Year C5 SS DS K=N K=L RDTYPE CHANGE
Lubricating oils and greases 1963 0.695 86247 82575 0.80 6.01 1 0
1968 0.716 93079 93079 1.28 10.93 1 0
1975 0.740 138734 135633 1.68 12.95 1 0
Blacks, mineral and carbon 1963 0.984 10142 8622 7.33 27.33 1 0
1968 0.971 12250 12250 8.60 33.66 1 0
Organic chemicals: acids (carboxylic), acid 1963 0.632 16869 15955 9.66 25.75 1 0
anhydrides, and acid halides 1968 0.648 23035 23035 16.29 49.23 1 0
Organic chemicals: alcohols (monohydric) 1963 0.946 51182 43596 9.66 25.75 2 0
1968 0.980 48608 48608 16.29 49.23 2 0
Organic chemicals: esters 1963 0.864 11326 9036 9.66 25.75 1 0
1968 0.825 18389 18389 16.29 49.23 1 0
Organic chemicals: halogenated derivatives of 1963 0.974 29000 25566 9.66 25.75 2 0
hydrocarbons 1968 0.969 43843 43843 16.29 49.23 2 0
1975 0.999 57312 73666 23.40 70.00 2 0
Organic chemicals: aliphatic hydrocarbons 1963 0.774 26560 20123 9.66 25.75 2 0
1968 0.993 39191 39191 16.29 49.23 2 0
Additives for liquid fuels and lubricating oils 1963 0.901 30588 26966 4.67 17.77 1 0
1968 0.807 41886 41886 6.48 25.04 1 0
1975 0.904 60214 62292 7.49 25.59 1 0
Bleaching materials 1963 0.816 14602 15129 4.67 17.77 1 0
1968 0.840 15215 15215 6.48 25.04 1 0

465
1975 0.880 12900 15281 7.49 25.59 1 0
Table B5 (continued)

466
Industry Year C5 SS DS K=N K=L RDTYPE CHANGE
Chemicals and preparations mainly for PVC 1963 0.773 19818 17453 4.67 17.77 1 0
processing 1968 0.871 24061 24061 6.48 25.04 1 0
1975 0.750 32430 33732 7.49 25.59 1 0
Food and perfume materials 1963 0.710 20616 18854 4.67 17.77 1 0
1968 0.811 26590 26590 6.48 25.04 1 0
1975 0.768 31028 30866 7.49 25.59 1 0
Rubber-processing chemicals and preparations 1963 0.985 13418 11250 4.67 17.77 1 0
1968 0.940 17118 17118 6.48 25.04 1 0
1975 0.979 16830 16585 7.49 25.59 1 0
Surface active materials 1963 0.668 13815 13960 4.67 17.77 1 0
1968 0.551 21260 21260 6.48 25.04 1 0
1975 0.687 23122 24279 7.49 25.59 1 0
Pharmaceutical chemicals 1963 0.539 52708 42369 1.68 4.93 2 0
1968 0.610 60624 60624 2.67 7.55 2 0
1975 0.660 67767 67286 4.73 11.40 2 0
Pharmaceutical preparations 1958 0.286 119871 89003 1.33 4.10 2 0
1963 0.287 168815 138163 1.68 4.93 2 0
1968 0.349 221326 221326 2.67 7.55 2 0
1975 0.393 341185 474269 4.73 11.40 2 0

Appendix B
Hair preparations 1958 0.670 16860 18822 0.41 2.02 2 0
1963 0.542 26510 27535 0.48 2.03 2 0
1968 0.570 31731 31731 0.92 2.96 2 0
1975 0.607 46032 60769 1.50 5.06 2 0
Dental preparations 1963 0.944 13587 12347 0.48 2.03 2 0

Data Sets
1968 0.874 14450 14450 0.92 2.96 2 0
1975 0.803 21586 24625 1.50 5.06 2 0
Other toilet preparations 1958 0.420 33779 37708 0.41 2.02 2 0
1963 0.373 50961 52931 0.48 2.03 2 0
1968 0.405 70585 70585 0.92 2.96 2 0
1975 0.421 99401 111633 1.50 5.06 2 0
Soaps 1958 0.832 59413 60862 2.27 7.35 1 0
1963 0.809 53272 53019 3.53 9.14 1 0
1968 0.801 48123 48123 4.44 14.43 1 0
1975 0.764 52683 49235 4.34 14.57 1 0
Finished detergents 1958 0.904 45542 36105 2.27 7.35 1 0
1963 0.845 55501 48629 3.53 9.14 1 0
1968 0.799 61265 61265 4.44 14.43 1 0
1975 0.828 90114 99083 4.34 14.57 1 0
Synthetic resins, thermosetting 1963 0.528 27759 20436 6.90 16.68 2 0
1968 0.487 40393 40393 7.90 19.17 2 0
Plastics materials other than synthetic resins, 1963 0.787 27581 24240 6.90 16.68 2 0
thermosetting 1968 0.768 30576 30576 7.90 19.17 2 0
1975 0.794 27783 31102 8.22 28.64 2 0
Plastics materials other than synthetic resins, 1963 0.745 32007 25633 6.90 16.68 2 0
thermoplastic 1968 0.665 54709 54709 7.90 19.17 2 0
Synthetic rubber 1958 0.917 2841 1641 5.12 12.18 2 0
1963 0.994 27581 21292 6.90 16.68 2 0
1968 0.994 37863 37863 7.90 19.17 2 0

467
1975 0.919 47882 40493 8.22 28.64 2 0
Table B5 (continued)

468
Industry Year C5 SS DS K=N K=L RDTYPE CHANGE
Finished synthetic organic dyestuffs 1958 0.971 29134 26262 4.79 9.73 2 0
1963 0.937 36627 36242 5.79 12.00 2 0
1968 0.904 40970 40970 10.71 20.56 2 0
1975 0.877 52043 59336 12.16 30.25 2 0
Polishes 1958 0.671 17337 18133 0.36 2.20 1 0
1963 0.796 17901 18436 0.47 3.04 1 0
1968 0.809 17249 17249 0.85 4.29 1 0
1975 0.849 14138 18195 1.31 7.31 1 0
Gelatine, glue, size, and other adhesives 1963 0.488 28641 29712 0.96 7.22 2 0
1968 0.461 36415 36415 1.39 7.99 2 0
1975 0.442 48031 44083 1.86 11.61 2 0
Explosives and fireworks 1958 0.897 47722 46778 1.77 3.58 1 0
1963 0.864 38202 37976 2.84 5.62 1 0
1968 0.835 36778 36778 4.09 7.59 1 0
Pesticides, disinfectants, and household 1958 0.424 29238 22616 0.41 2.96 2 0
deodorisers 1963 0.520 34655 31491 0.68 4.56 2 0
1968 0.534 44697 44697 1.28 7.49 2 0
1975 0.533 73971 80699 2.57 13.13 2 0
Welding and flame-cutting machines 1963 0.638 11802 11638 0.77 3.26 1 0

Appendix B
1968 0.501 14590 14590 0.96 4.02 1 0
1975 0.630 18124 15254 0.87 4.85 1 0
Air and gas compressors and exhausters 1958 0.415 17780 17914 0.63 2.37 1 1

Data Sets
1963 0.481 24848 24502 0.71 2.72 1 1
1968 0.572 33439 33439 0.77 2.89 1 1
1975 0.719 36671 33365 1.03 3.98 1 1
Internal combustion engines, industrial 1958 0.740 44160 39449 2.79 4.07 1 1
1963 0.677 55144 58265 3.02 4.11 1 1
1975 0.889 85484 79010 6.04 7.22 1 1
Textile machinery 1958 0.647 51270 51656 0.53 2.42 1 0
1963 0.603 73907 72878 0.66 2.59 1 0
1968 0.592 101346 101346 0.72 2.97 1 0
1975 0.619 93273 88189 0.95 4.15 1 0
Powered industrial trucks and industrial tractors 1958 0.698 9424 10001 0.29 1.41 1 0
1963 0.764 16297 15212 0.33 1.56 1 0
1968 0.555 42182 42182 0.41 1.91 1 0
1975 0.607 71547 66984 0.53 2.44 1 0
Printing, bookbinding, paperworking, etc. 1958 0.446 20260 20413 0.38 2.15 1 0
machinery 1963 0.491 28556 28159 0.49 2.43 1 0
1968 0.491 33563 33563 0.67 2.98 1 0
1975 0.440 39662 37500 0.75 3.55 1 0
Rubber and plastics working machinery 1963 0.563 16686 16453 0.49 2.23 1 0
1968 0.506 24158 24158 0.59 2.82 1 0
1975 0.508 19540 19228 0.71 3.84 1 0
Photographic and cinematographic apparatus and 1963 0.706 14917 15978 0.54 3.10 2 0
appliances, and document copying equipment 1968 0.843 38812 38812 1.02 2.87 2 0

469
1975 0.787 54097 63083 1.78 7.28 2 0
Table B5 (continued)

470
Industry Year C5 SS DS K=N K=L RDTYPE CHANGE
Optical instruments and appliances 1963 0.507 9436 10108 0.36 1.35 2 0
1968 0.570 14143 14143 0.51 1.86 2 0
1975 0.572 20023 23349 0.68 3.11 2 0
Transformers for lighting, heating, and power 1958 0.499 43892 38349 1.12 1.87 2 1
1963 0.450 60208 55061 1.10 1.98 2 1
1968 0.767 55120 55120 1.28 3.04 2 1
1975 0.689 38395 38395 1.28 3.78 2 1
Starting and controlling gear for electric motors 1958 0.613 33487 29258 1.12 1.87 1 1
1963 0.479 39976 36558 1.10 1.98 1 1
1968 0.526 44871 44871 1.28 3.04 1 1
1975 0.556 46821 46821 1.28 3.78 1 1
Switchgear and switchboards 1958 0.616 79351 69329 1.12 1.87 2 1
1963 0.609 96849 88570 1.10 1.98 2 1
1968 0.774 100001 100001 1.28 3.04 2 1
1975 0.664 70307 70307 1.28 3.78 2 1
Line apparatus for long-distance communication 1958 0.939 11364 10457 2.04 2.08 2 1
1963 0.962 23632 22753 3.22 2.81 2 1
1968 1.000 40143 40143 2.77 3.28 2 1
1975 0.949 31233 34038 3.39 3.62 2 1

Appendix B
Passive components 1963 0.353 76422 73580 0.60 1.63 1 0
1968 0.335 125370 125370 0.85 2.40 1 0
1975 0.355 130312 133557 1.44 4.39 1 0
Television receiving sets 1958 0.528 78964 81098 0.53 1.10 1 0

Data Sets
1963 0.816 66324 72215 0.55 1.00 1 0
1968 0.927 86103 86103 1.01 1.98 1 0
1975 0.850 135769 201970 1.22 2.80 1 0
Radio receiving sets 1963 0.579 23118 25140 0.55 1.00 1 0
1968 0.747 11556 11556 1.01 1.98 1 0
1975 0.835 4939 7389 1.22 2.80 1 0
Gramophones, record players, and tape recorders 1958 0.680 29788 25884 0.53 1.10 1 0
1963 0.553 35794 34463 0.55 1.00 1 0
Electronic computers 1963 0.733 46845 45102 1.80 4.68 2 0
1968 0.872 117633 117633 2.30 4.22 2 0
1975 0.864 213692 232886 2.38 5.18 2 0
Radio communication equipment 1958 0.646 14442 13289 1.56 3.00 2 0
1963 0.638 38064 36648 1.37 2.80 2 0
1968 0.808 44742 44742 1.37 3.32 2 0
1975 0.894 61650 67188 1.55 3.75 2 0
Radar and electronic navigation aid equipment 1958 0.783 21413 19704 1.56 3.00 2 0
1963 0.722 64096 61712 1.37 2.80 2 0
1968 0.746 82918 82918 1.37 3.32 2 0
1975 0.807 86093 93826 1.55 3.75 2 0
Electronic measuring and testing instruments 1958 0.467 11790 10849 1.56 3.00 2 0
1963 0.369 18071 17399 1.37 2.80 2 0
Washing machines, electrically operated 1958 0.764 34924 25410 1.18 2.55 1 1
1963 0.852 55575 44969 1.18 2.27 1 1
1968 0.869 36915 36915 1.43 3.00 1 1

471
1975 0.981 44509 52813 2.23 4.79 1 1
Table B5 (continued)

472
Industry Year C5 SS DS K=N K=L RDTYPE CHANGE
Electrical equipment for motor vehicles, cycles, 1958 0.747 83330 89751 0.49 1.72 1 0
and aircraft, excluding accumulators 1963 0.701 98189 97269 0.63 2.03 1 0
1968 0.740 125913 125913 0.98 2.68 1 0
1975 0.807 123426 134512 1.43 4.13 1 0
Batteries and accumulators 1958 0.782 42602 48030 0.49 1.72 1 1
1963 0.787 52821 56933 0.63 2.03 1 1
1968 0.841 66757 66757 0.98 2.68 1 1
1975 0.908 74751 76167 1.43 4.13 1 1
Electric lamps 1958 0.744 23666 18932 0.49 1.72 2 1
1963 0.717 35084 29795 0.63 2.03 2 1
1968 0.866 40308 40308 0.98 2.68 2 1
1975 0.941 35415 41413 1.43 4.13 2 1
Marine machinery 1963 0.346 56700 49781 0.86 1.96 1 0
1968 0.574 60135 60135 0.98 2.11 1 0
1975 0.457 55276 56078 1.36 2.89 1 0
Cars 1958 0.901 521809 471432 3.47 6.46 1 0
1963 0.912 752514 685258 3.49 5.81 1 0
1968 0.992 853240 853240 3.71 6.49 1 0
1975 0.982 690957 648622 3.78 6.78 1 0

Appendix B
Internal combustion engines for motor vehicles 1963 0.931 80452 78340 3.49 5.81 1 0
1968 0.970 85532 85532 3.71 6.49 1 0
1975 0.914 48412 45773 3.78 6.78 1 0
Aero-engines 1958 0.940 152178 157306 2.24 2.32 2 0

Data Sets
1963 0.971 108523 107012 3.00 2.58 2 0
Cutlery 1958 0.658 22897 20524 0.24 1.75 1 0
1963 0.669 29222 26752 0.39 2.20 1 0
1968 0.709 33214 33214 0.56 3.42 1 0
1975 0.560 25796 27019 0.60 4.41 1 0
Tyres and tubes, other than retreaded tyres 1958 0.932 112412 94782 1.27 3.80 1 1
1963 0.945 144054 130074 1.54 4.06 1 1
1968 0.928 184836 184836 2.39 5.90 1 1
1975 0.964 190550 199087 2.68 7.51 1 1
Notes: C5 is the five-firm sales concentration ratio.
SS is sales revenue (in £1000) deflated by the general producer price index.
DS is sales revenue (in £1000) deflated by industry-specific producer price indices.
K is the value of capital stock of the corresponding MLH industry in 1980 prices (in £ million).
L is employment of the corresponding MLH industry (in 1000).
N is the number of plants with at least 25 employees of the corresponding MLH industry.
RDTYPE takes the value 1 for an industry with typical R&D-sales ratio between 1% and 2%, and 2 for an industry with typical R&D-sales ratio
higher than 2%.
CHANGE takes the value 1 for an industry with a change in competition regime and 0 otherwise.

473
Table B6

474
Data set for chapter 7
Industry Year NFIRMS NPLANTS SS PROFIT PCM K/N K/L UNION ADTYPE RDTYPE CHANGE
Slate and slate products 1954 16 17 2601 498 0.192 0.15 0.64 0.455 0 0 1
1958 17 18 2249 477 0.212 0.13 0.69 0.408 0 0 1
1963 16 21 2084 467 0.224 0.12 1.17 0.464 0 0 1
1968 15 17 2169 528 0.243 0.21 2.35 0.395 0 0 1
Clay, brick earth, marl, shale, 1954 48 116 9927 3185 0.321 0.26 4.67 0.455 0 0 1
and chalk 1958 39 120 10955 3835 0.350 0.31 5.91 0.408 0 0 1
1963 24 87 16964 5632 0.332 0.63 7.74 0.464 0 0 1
1968 20 102 30359 15311 0.504 1.01 12.61 0.395 0 0 1
Iron ore and ironstone 1954 19 46 10093 3082 0.305 1.18 7.64 0.455 0 0 0
1958 19 44 11714 3976 0.339 1.55 9.78 0.408 0 0 0
1963 11 50 11162 4534 0.406 1.26 11.51 0.464 0 0 0
Salt 1954 8 23 14395 4177 0.290 1.36 6.31 0.455 0 0 0
1958 8 19 14449 5913 0.409 1.94 8.50 0.408 0 0 0
1963 6 18 18729 10628 0.567 3.17 15.14 0.464 0 0 0
Nonmetalliferous mining and 1954 18 45 8771 1589 0.181 0.41 3.39 0.455 0 0 0
quarrying other than salt 1958 20 40 7088 1015 0.143 0.45 4.38 0.408 0 0 0
1963 13 24 5197 1859 0.358 0.73 7.77 0.464 0 0 0
Grain milling (MLH 211) 1954 149 227 322879 30864 0.096 0.53 4.16 0.305 0 0 1

Appendix B
1958 134 223 306594 44288 0.144 0.65 4.82 0.283 0 0 1
1963 78 183 278566 46326 0.166 1.04 7.17 0.295 0 0 1
1968 61 149 283599 54094 0.191 1.50 10.38 0.346 0 0 1
1973 56 133 275183 44925 0.163 1.87 13.23 0.490 0 0 1
Bread and flour confectionery 1954 763 1096 241093 34421 0.143 0.32 3.17 0.305 0 0 1

Data Sets
(MLH 212) 1958 554 839 293119 45856 0.156 0.43 2.86 0.283 0 0 1
1963 271 728 364153 62594 0.172 0.54 2.58 0.295 0 0 1
1968 209 536 381206 71191 0.187 0.84 3.03 0.346 0 0 1
1973 250 507 421962 77066 0.183 0.98 3.30 0.490 0 0 1
Biscuits (MLH 213) 1954 77 111 115973 20859 0.180 0.88 1.76 0.305 2 0 1
1958 60 95 122312 21270 0.174 1.51 2.65 0.283 2 0 1
1963 37 75 133484 31649 0.237 2.28 3.74 0.295 2 0 1
1968 33 65 149081 35662 0.239 3.22 4.39 0.346 2 0 1
1973 29 54 149029 26332 0.177 4.42 5.37 0.490 2 0 1
Bacon and ham (cured and 1954 146 185 176828 10066 0.057 0.26 2.21 0.305 0 0 1
smoked), sausages and sausage 1958 134 168 146719 14883 0.101 0.33 2.49 0.283 0 0 1
meat 1963 90 138 157864 14325 0.091 0.50 2.08 0.295 0 0 1
1968 67 101 166037 18246 0.110 0.80 2.49 0.346 0 0 1
Preserved meat or fish, meat 1963 50 74 43787 11967 0.273 0.23 1.57 0.295 1 0 0
extracts, fish cured, smoked, or 1968 43 65 59556 13788 0.232 0.59 2.89 0.346 1 0 0
salted, etc.
Other meat products, offal, lard, 1963 103 173 67826 8308 0.122 0.17 2.22 0.295 0 0 0
etc. 1968 99 130 74425 11179 0.150 0.35 2.85 0.346 0 0 0
Butter 1954 27 46 47012 2728 0.058 0.99 13.15 0.305 0 0 1
1958 26 42 60707 3308 0.054 1.26 12.16 0.283 0 0 1
1963 15 33 59916 2701 0.045 1.61 15.83 0.295 0 0 1
1968 17 38 74121 4327 0.058 1.59 16.08 0.346 0 0 1

475
Table B6 (continued)

476
Industry Year NFIRMS NPLANTS SS PROFIT PCM K/N K/L UNION ADTYPE RDTYPE CHANGE
Condensed milk 1954 9 16 24685 1553 0.063 1.67 9.39 0.305 2 0 1
1958 7 19 36241 2871 0.079 1.73 8.85 0.283 2 0 1
1963 5 14 31741 5205 0.164 3.37 16.58 0.295 2 0 1
1968 6 10 29158 4758 0.163 5.86 21.74 0.346 2 0 1
Milk powder 1954 9 15 15664 913 0.058 0.73 7.54 0.305 0 0 1
1958 5 9 12789 648 0.051 0.94 8.78 0.283 0 0 1
1963 3 5 7801 887 0.114 2.42 16.35 0.295 0 0 1
Cream and other milk products 1954 14 22 20159 2894 0.144 0.97 6.62 0.305 1 0 1
(1958 S.I.C.) 1958 12 21 19366 3469 0.179 1.25 8.28 0.283 1 0 1
1963 15 27 27786 3516 0.127 1.56 9.19 0.295 1 0 1
Ice cream 1954 22 28 20749 5140 0.248 1.65 8.97 0.305 2 0 0
1958 14 22 28054 8143 0.290 4.18 12.27 0.283 2 0 0
1963 13 22 19029 4039 0.212 3.97 20.86 0.295 2 0 0
Cocoa and chocolate products 1954 58 79 185400 26256 0.142 1.28 2.01 0.305 2 0 1
1958 53 75 174518 32179 0.184 2.10 2.85 0.283 2 0 1
1963 41 65 175872 43085 0.245 3.63 4.20 0.295 2 0 1
1968 30 50 195358 43690 0.224 5.79 5.88 0.346 2 0 1
Fruit and vegetable products 1954 209 309 152109 21938 0.144 0.29 1.65 0.305 2 0 0
(MLH 218) 1958 168 263 190250 30516 0.160 0.44 1.97 0.283 2 0 0

Appendix B
1963 131 223 248688 51160 0.206 0.88 3.13 0.295 2 0 0
1968 119 201 286069 60947 0.213 1.29 4.41 0.346 2 0 0
1973 110 172 369177 76929 0.208 1.96 5.34 0.490 2 0 0
Dog and cat foods 1954 10 12 9937 2148 0.216 0.43 2.53 0.305 2 0 0

Data Sets
1958 12 14 20516 4363 0.213 0.58 2.72 0.283 2 0 0
1963 8 15 30693 9976 0.325 1.05 4.71 0.295 2 0 0
1968 12 17 44694 9639 0.216 1.67 6.61 0.346 2 0 0
Vegetable and animal oils and 1954 47 74 131158 12767 0.097 0.74 5.15 0.330 0 0 0
fats (MLH 221) 1958 44 76 158533 11888 0.075 0.89 6.15 0.311 0 0 0
1963 45 76 127451 13258 0.104 1.02 8.43 0.334 0 0 0
1968 45 77 113298 12776 0.113 1.24 12.30 0.394 0 0 0
1973 41 65 173956 23076 0.133 1.95 16.13 0.471 0 0 0
Margarine (MLH 229.1) 1954 17 18 68701 6919 0.101 1.73 5.85 0.305 2 0 0
1958 12 14 62065 6364 0.103 2.81 7.83 0.283 2 0 0
1963 8 9 52338 3932 0.075 4.47 9.86 0.295 2 0 0
1968 7 8 40621 2240 0.055 5.43 10.76 0.346 2 0 0
1973 7 8 59649 5270 0.088 7.18 14.96 0.490 2 0 0
Coffee, and coffee and chicory 1954 7 9 11152 1192 0.107 0.47 4.45 0.305 2 0 0
extracts and essences 1958 7 10 21788 3821 0.175 1.16 6.30 0.283 2 0 0
1963 9 13 29465 6999 0.238 1.58 6.44 0.295 2 0 0
Spirits, distilled 1958 16 65 85846 4758 0.055 0.34 6.20 0.283 0 0 1
1963 16 70 72550 6366 0.088 0.67 11.62 0.295 0 0 1
1968 17 74 118681 9181 0.077 0.89 15.39 0.346 0 0 1
British wines, cider, and perry 1954 18 30 13693 3312 0.242 0.54 4.78 0.305 2 0 0
(MLH 239.2) 1958 18 26 18830 6327 0.336 0.67 4.15 0.283 2 0 0
1963 7 17 19466 6030 0.310 1.20 5.21 0.295 2 0 0
1968 10 19 27750 6841 0.247 1.32 5.85 0.346 2 0 0
1973 9 17 41139 13033 0.317 2.02 7.23 0.490 2 0 0

477
Table B6 (continued)

478
Industry Year NFIRMS NPLANTS SS PROFIT PCM K/N K/L UNION ADTYPE RDTYPE CHANGE
Tobacco (MLH 240) 1958 26 62 1053944 65865 0.062 2.24 3.14 0.570 2 0 0
1963 16 47 1224000 79042 0.065 4.46 4.86 0.673 2 0 0
1968 12 41 1255888 100639 0.080 7.66 7.73 0.769 2 0 0
1973 10 34 1134385 117867 0.104 11.65 10.10 0.994 2 0 0
Coke and manufactured fuel 1954 28 92 166172 15091 0.091 3.24 14.14 0.330 0 0 0
(MLH 261) 1958 25 83 214270 22461 0.105 6.02 21.60 0.311 0 0 0
1963 24 56 198204 12943 0.065 9.21 28.43 0.334 0 0 0
Lubricating oils and greases 1954 54 82 55148 13252 0.240 0.46 5.15 0.330 1 1 0
(MLH 263) 1958 46 73 59879 15703 0.262 0.62 5.92 0.311 1 1 0
1963 40 67 57484 16006 0.278 0.72 7.31 0.334 1 1 0
1968 37 61 58883 13796 0.234 1.25 12.75 0.394 1 1 0
1973 34 54 83814 17366 0.207 1.57 14.06 0.471 1 1 0
Miscellaneous basic chemicals 1958 96 182 174748 60730 0.348 5.51 23.47 0.311 0 1 0
(1958 S.I.C.) 1963 104 207 263477 99773 0.379 6.44 32.45 0.334 0 1 0
General chemicals, other than 1963 97 141 171835 57316 0.334 4.11 18.87 0.334 0 1 0
inorganic and organic (MLH 1968 88 127 211793 62688 0.296 6.27 27.19 0.394 0 1 0
271.3, 1968 S.I.C.) 1973 91 132 341297 83509 0.245 7.36 28.99 0.471 0 1 0
Pharmaceutical chemicals and 1954 152 184 136436 45331 0.332 1.00 4.25 0.330 2 2 0
preparations (MLH 272) 1958 130 160 169296 57592 0.340 1.45 4.57 0.311 2 2 0

Appendix B
1963 129 182 227398 89222 0.392 1.64 5.09 0.334 2 2 0
1968 103 150 279096 111350 0.399 2.81 7.92 0.394 2 2 0
1973 117 148 414171 153658 0.371 4.70 11.86 0.471 2 2 0
Toilet preparations (MLH 273) 1954 57 60 37424 15031 0.402 0.37 2.03 0.330 2 2 0

Data Sets
1958 59 64 48508 18264 0.377 0.41 2.09 0.311 2 2 0
1963 69 77 73365 31103 0.424 0.50 2.19 0.334 2 2 0
1968 56 67 100269 38634 0.385 1.03 3.17 0.394 2 2 0
1973 69 69 137667 54393 0.395 1.47 4.58 0.471 2 2 0
Paint (MLH 274) 1954 198 272 128395 29220 0.228 0.39 3.04 0.330 2 1 1
1958 173 248 141652 32847 0.232 0.53 3.47 0.311 2 1 1
1963 136 219 148773 38306 0.257 0.64 3.93 0.334 2 1 1
1968 96 146 143872 39135 0.272 0.97 4.81 0.394 2 1 1
1973 99 131 173887 54041 0.311 1.25 6.07 0.471 2 1 1
Soap and detergents (MLH 275) 1954 66 82 109731 20591 0.188 1.84 7.24 0.330 2 1 0
1958 56 73 125164 27885 0.223 2.27 7.91 0.311 2 1 0
1963 42 55 134478 35697 0.265 3.66 9.78 0.334 2 1 0
1968 37 48 128781 32399 0.252 4.81 16.03 0.394 2 1 0
1973 45 58 146130 32968 0.226 4.14 17.43 0.471 2 1 0
Synthetic resins and plastics 1954 — 63 94210 21470 0.228 4.29 12.88 0.330 0 2 0
materials 1958 57 78 146874 28741 0.196 4.38 11.56 0.311 0 2 0
1963 62 101 194080 46757 0.241 5.08 15.96 0.334 0 2 0
1968 85 128 324191 82940 0.256 6.92 19.15 0.394 0 2 0
Synthetic rubber 1963 4 4 22560 6632 0.294 27.28 72.01 0.334 0 2 0
1968 4 5 34617 10825 0.313 24.18 62.64 0.394 0 2 0
Dyestuffs 1954 18 28 73182 15085 0.206 4.21 6.13 0.330 0 2 0
1958 15 23 58603 13441 0.229 5.10 6.77 0.311 0 2 0
1963 15 27 71993 18408 0.256 6.56 10.52 0.334 0 2 0
1968 15 20 79126 22532 0.285 14.04 17.77 0.394 0 2 0

479
Table B6 (continued)

480
Industry Year NFIRMS NPLANTS SS PROFIT PCM K/N K/L UNION ADTYPE RDTYPE CHANGE
Polishes (MLH 279.1) 1954 38 50 19930 4925 0.247 0.20 1.93 0.330 2 1 0
1958 30 38 21543 6875 0.319 0.35 2.51 0.311 2 1 0
1963 34 46 26856 7891 0.294 0.46 3.35 0.334 2 1 0
1968 29 37 33787 10921 0.323 0.80 4.70 0.394 2 1 0
1973 29 33 38620 12772 0.331 1.10 6.52 0.471 2 1 0
Formulated adhesives, gelatine, 1963 36 55 26219 5894 0.225 0.84 7.86 0.334 0 2 0
glue, size, etc. (MLH 279.2) 1968 29 42 39693 9217 0.232 1.39 9.00 0.394 0 2 0
1973 25 36 44826 11298 0.252 1.98 12.78 0.471 0 2 0
Fireworks 1954 17 21 3280 989 0.302 0.04 0.32 0.330 0 0 1
1958 15 20 3646 857 0.235 0.06 0.41 0.311 0 0 1
1963 12 20 3554 1029 0.290 0.11 0.89 0.334 0 0 1
Explosives 1954 9 42 65707 12172 0.185 1.59 1.60 0.330 0 1 0
1958 9 37 51972 12527 0.241 2.98 3.94 0.311 0 1 0
1963 9 20 41663 12129 0.291 5.63 6.28 0.334 0 1 0
1968 6 12 39978 11432 0.286 10.30 8.25 0.394 0 1 0
Formulated pesticides and 1954 35 42 18211 3791 0.208 0.31 2.56 0.330 2 2 0
disinfectants (MLH 279.4) 1958 39 49 22835 5787 0.253 0.34 2.96 0.311 2 2 0
1963 30 40 27686 8439 0.305 0.62 4.83 0.334 2 2 0
1968 25 33 32399 7832 0.242 1.20 8.14 0.394 2 2 0

Appendix B
1973 22 22 65472 19064 0.291 2.44 16.21 0.471 2 2 0
Surgical bandages, etc. and 1963 22 42 35148 11317 0.322 1.13 4.04 0.562 0 0 1
sanitary towels (MLH 279.6) 1968 18 37 42561 14561 0.342 2.00 6.11 0.602 0 0 1
1973 16 34 53802 16919 0.314 2.32 6.30 0.722 0 0 1
Steel forgings 1958 23 29 13645 2756 0.202 0.52 3.95 0.499 0 0 1

Data Sets
1963 26 35 24829 3362 0.135 0.92 7.05 0.528 0 0 1
Steel sheets and tinplate 1958 17 34 174834 11601 0.066 4.97 7.54 0.499 0 0 1
1963 11 19 105977 8763 0.083 7.08 12.96 0.528 0 0 1
Steel manufacture without 1958 130 184 259280 21407 0.083 0.90 4.13 0.499 0 0 0
melting, other than forgings, 1963 99 154 294025 25729 0.088 2.29 7.28 0.528 0 0 0
sheets, and tinplate
Steel castings 1958 52 63 45413 9411 0.207 0.84 2.42 0.499 0 0 1
1963 51 66 47703 9490 0.199 1.19 3.75 0.528 0 0 1
Steel manufacture with melting, 1958 48 95 897286 142228 0.159 28.73 14.82 0.499 0 0 0
other than castings 1963 41 97 822988 127014 0.154 41.75 22.42 0.528 0 0 0
Steel tubing and gas cylinders 1958 22 55 189335 33744 0.178 4.53 5.67 0.499 0 0 1
1963 23 61 187372 28462 0.152 6.25 8.44 0.528 0 0 1
1968 30 70 209559 39712 0.190 6.09 9.10 0.567 0 0 1
Steel tubes, manipulated, 1958 42 63 23553 4069 0.173 0.25 1.81 0.499 0 0 1
fabricated, etc. 1963 52 69 31605 6555 0.207 0.38 2.71 0.528 0 0 1
1968 39 53 34377 7230 0.210 0.71 3.81 0.567 0 0 1
Pig iron 1958 24 24 55198 5607 0.102 2.28 7.12 0.499 0 0 0
1963 10 13 37820 4241 0.112 4.52 12.46 0.528 0 0 0
Pressure pipes and fittings 1958 21 32 39405 5175 0.131 3.37 7.90 0.499 0 0 1
1963 15 25 38602 5820 0.151 4.02 8.23 0.528 0 0 1
1968 11 23 38671 3970 0.103 3.46 5.90 0.567 0 0 1
Iron castings for the building 1958 53 71 32110 6252 0.195 0.20 0.89 0.499 0 0 1
industry 1963 39 57 37901 9914 0.262 0.51 2.02 0.528 0 0 1

481
1968 29 42 35835 8025 0.224 1.27 4.22 0.567 0 0 1
Table B6 (continued)

482
Industry Year NFIRMS NPLANTS SS PROFIT PCM K/N K/L UNION ADTYPE RDTYPE CHANGE
Ingot moulds and bottoms 1958 19 19 16357 2386 0.146 2.17 10.37 0.499 0 0 1
1963 13 14 15561 3217 0.207 2.64 10.38 0.528 0 0 1
1968 9 17 18545 3370 0.182 1.86 7.74 0.567 0 0 1
Marine and other engineering 1958 321 341 58702 8119 0.138 0.17 1.71 0.499 0 0 1
castings 1963 258 278 54862 9124 0.166 0.32 3.12 0.528 0 0 1
1968 183 208 48501 9335 0.192 0.49 4.57 0.567 0 0 1
Aluminium and aluminium 1954 23 36 98896 10417 0.105 3.64 4.39 0.509 0 0 1
alloys: fabricated 1958 27 44 106869 12093 0.113 4.00 5.46 0.499 0 0 1
1963 35 63 145189 19994 0.138 3.68 6.44 0.528 0 0 1
1968 38 62 171511 20356 0.119 4.95 9.26 0.567 0 0 1
Aluminium and aluminium 1954 76 91 30411 3534 0.116 0.28 1.91 0.509 0 0 1
alloys: founded 1958 74 91 31342 4292 0.137 0.35 2.39 0.499 0 0 1
1963 90 115 41397 7470 0.180 0.46 3.07 0.528 0 0 1
1968 90 120 57766 9736 0.169 0.89 5.17 0.567 0 0 1
Copper, brass and other copper 1954 24 28 64971 2072 0.032 0.56 4.92 0.509 0 0 1
alloys: smelted and refined 1958 23 27 42234 2209 0.052 0.51 5.90 0.499 0 0 1
1963 31 42 52461 3415 0.065 0.36 3.59 0.528 0 0 1
1968 20 26 105851 5909 0.056 1.36 10.28 0.567 0 0 1
Copper, brass and other copper 1954 169 207 227902 24357 0.107 1.03 4.82 0.509 0 0 1

Appendix B
alloys: fabricated and founded 1958 136 179 202096 23157 0.115 1.35 5.47 0.499 0 0 1
1963 126 176 259488 33028 0.127 1.55 6.41 0.528 0 0 1
1968 116 168 342383 37919 0.111 1.73 7.06 0.567 0 0 1
Miscellaneous nonferrous metals 1954 89 123 165314 16254 0.098 1.40 7.54 0.509 0 0 0

Data Sets
(MLH 323) 1958 97 127 175503 16495 0.094 1.52 7.04 0.499 0 0 0
1963 79 123 199829 22709 0.114 1.74 7.59 0.528 0 0 0
1968 80 115 310325 32855 0.106 2.11 10.34 0.567 0 0 0
1973 78 107 283136 42038 0.148 2.60 15.45 0.681 0 0 0
Welding and flame-cutting 1954 29 35 17504 2657 0.152 0.74 4.38 0.509 0 1 0
equipment 1958 31 40 23903 4203 0.176 0.76 4.04 0.499 0 1 0
1963 31 36 18480 3443 0.186 0.60 3.49 0.528 0 1 0
1968 33 44 32371 6141 0.190 0.48 2.65 0.567 0 1 0
Industrial engines (MLH 334) 1954 41 58 70673 13542 0.192 2.24 4.00 0.509 0 1 1
1958 35 55 80071 12064 0.151 2.89 4.17 0.499 0 1 1
1963 27 50 94382 13903 0.147 3.08 4.17 0.528 0 1 1
1968 21 44 115623 23716 0.205 3.61 4.77 0.567 0 1 1
1973 16 30 107459 20053 0.187 5.27 6.12 0.681 0 1 1
Textile machinery and 1954 266 338 95689 21590 0.226 0.36 1.99 0.509 0 1 0
accessories (MLH 335) 1958 189 245 77793 14165 0.182 0.52 2.69 0.499 0 1 0
1963 148 204 113738 28142 0.247 0.65 2.83 0.528 0 1 0
1968 146 213 140642 35674 0.254 0.69 3.22 0.567 0 1 0
1973 123 176 143124 39012 0.273 0.91 4.31 0.681 0 1 0
Lifts and escalators 1954 32 54 10618 2835 0.267 0.04 0.36 0.509 0 0 1
1958 30 62 15773 3688 0.234 0.05 0.39 0.499 0 0 1
1963 32 70 29641 7099 0.239 0.18 1.10 0.528 0 0 1
1968 28 62 38827 11528 0.297 0.33 1.65 0.567 0 0 1

483
Table B6 (continued)

484
Industry Year NFIRMS NPLANTS SS PROFIT PCM K/N K/L UNION ADTYPE RDTYPE CHANGE
Powered industrial trucks and 1954 7 9 3884 211 0.054 0.19 1.27 0.509 1 1 0
industrial tractors 1958 6 6 7055 981 0.139 0.32 0.86 0.499 1 1 0
1963 17 23 19281 3175 0.165 0.22 0.92 0.528 1 1 0
1968 28 35 36516 9248 0.253 0.51 2.15 0.567 1 1 0
Mining machinery (MLH 339.1) 1954 53 66 41508 8456 0.204 0.50 2.13 0.509 0 0 0
1958 67 80 60392 11032 0.183 0.52 2.12 0.499 0 0 0
1963 66 84 68058 15533 0.228 0.57 2.44 0.528 0 0 0
1968 63 91 85380 16290 0.191 0.70 2.91 0.567 0 0 0
1973 56 75 99856 22153 0.222 0.97 3.64 0.681 0 0 0
Printing, bookbinding, and 1954 81 100 26214 7275 0.278 0.32 1.96 0.509 0 1 0
paper goods making machinery 1958 71 86 26670 7129 0.267 0.39 2.15 0.499 0 1 0
(MLH 339.2) 1963 66 93 35247 10502 0.298 0.48 2.59 0.528 0 1 0
1968 71 103 54335 12159 0.224 0.63 3.18 0.567 0 1 0
1973 72 79 62216 12899 0.207 0.87 4.37 0.681 0 1 0
Space-heating, ventilating and 1954 125 148 30226 6520 0.216 0.13 1.15 0.509 0 0 0
air-conditioning equipment 1958 121 136 39389 7976 0.203 0.15 1.11 0.499 0 0 0
(MLH 339.4)
1963 129 167 64507 16584 0.257 0.25 1.75 0.528 0 0 0
1968 171 221 96991 24068 0.248 0.38 2.80 0.567 0 0 0
Scales and weighing machinery 1954 17 45 9621 2682 0.279 0.23 1.88 0.509 0 0 0

Appendix B
1958 15 44 8447 2010 0.238 0.22 1.77 0.499 0 0 0
1963 14 50 11828 3291 0.278 0.23 1.93 0.528 0 0 0
1968 18 29 14481 4293 0.296 0.42 2.24 0.567 0 0 0
Pulpmaking and papermaking 1954 11 14 7933 1701 0.214 0.61 2.93 0.509 0 0 0

Data Sets
machinery 1958 16 19 17846 3427 0.192 0.73 2.59 0.499 0 0 0
1963 11 17 24607 4360 0.177 1.25 4.59 0.528 0 0 0
Packaging and bottling 1963 37 52 19727 4576 0.232 0.27 1.31 0.528 0 0 0
machinery 1968 47 63 29047 7010 0.241 0.67 3.58 0.567 0 0 0
Constructional and other 1963 292 359 139771 26769 0.192 0.37 2.70 0.528 0 0 1
fabricated ironwork and 1968 406 493 193201 39907 0.207 0.33 2.66 0.567 0 0 1
steelwork
Ordnance and small arms (MLH 1958 42 75 69364 11952 0.172 1.50 2.78 0.499 0 0 0
342) 1963 32 46 54281 10584 0.195 2.36 3.49 0.528 0 0 0
1968 30 40 54619 13273 0.243 2.92 5.16 0.567 0 0 0
1973 22 35 63398 8966 0.141 4.11 7.30 0.681 0 0 0
Photographic and document 1954 29 36 6965 2032 0.292 0.39 3.63 0.509 2 2 0
copying equipment (MLH 351) 1958 16 19 6710 1886 0.281 0.73 4.11 0.499 2 2 0
1963 22 25 10814 2433 0.225 0.52 3.62 0.528 2 2 0
1968 19 25 40611 8106 0.200 1.10 3.19 0.567 2 2 0
1973 25 35 50657 8313 0.164 1.59 6.27 0.681 2 2 0
Watches and parts of watches 1954 8 13 6157 1364 0.222 1.20 3.49 0.509 2 0 0
1958 10 12 5312 687 0.129 1.08 3.41 0.499 2 0 0
1963 7 9 5944 1425 0.240 1.68 4.02 0.528 2 0 0
1968 10 15 11373 1330 0.117 1.89 4.32 0.567 2 0 0
Spectacles and parts of 1954 82 129 13735 2438 0.178 0.07 0.89 0.509 0 1 1
spectacles 1958 78 117 14775 3497 0.237 0.08 0.96 0.499 0 1 1
1963 76 123 17241 3081 0.179 0.09 1.13 0.528 0 1 1
1968 68 97 19438 4225 0.217 0.15 1.58 0.567 0 1 1

485
Table B6 (continued)

486
Industry Year NFIRMS NPLANTS SS PROFIT PCM K/N K/L UNION ADTYPE RDTYPE CHANGE
Optical instruments 1954 21 30 9597 2589 0.270 0.39 1.42 0.509 0 2 0
1958 24 36 7776 1994 0.256 0.28 1.83 0.499 0 2 0
1963 27 43 14877 3646 0.245 0.33 1.49 0.528 0 2 0
Scientific instruments, other than 1954 169 237 70464 17754 0.252 0.31 1.58 0.509 0 2 0
optical instruments 1958 175 246 88353 22821 0.258 0.35 1.59 0.499 0 2 0
1963 171 253 125441 32006 0.255 0.51 2.09 0.528 0 2 0
Scientific and industrial 1963 247 417 200021 52256 0.261 0.34 1.42 0.528 0 2 0
instruments and systems 1968 280 464 285892 79626 0.279 0.49 1.96 0.567 0 2 0
(MLH 354)
1973 332 461 267226 67153 0.251 0.65 3.29 0.681 0 2 0
Electrical machinery (MLH 361) 1954 232 353 362912 70609 0.195 0.93 1.72 0.509 0 2 1
1958 228 373 449783 99582 0.221 1.09 1.92 0.499 0 2 1
1963 234 393 477502 75169 0.157 1.11 2.03 0.528 0 2 1
1968 220 401 485879 116319 0.239 1.27 3.13 0.567 0 2 1
1973 255 388 476458 93072 0.195 1.34 3.87 0.681 0 2 1
Insulated wires and cables 1954 50 83 163215 17718 0.109 1.95 3.45 0.509 0 1 1
(MLH 362) 1958 42 70 167783 20815 0.124 2.79 4.30 0.499 0 1 1
1963 47 91 226293 38766 0.171 2.72 4.69 0.528 0 1 1
1968 32 78 313794 46708 0.149 3.75 5.47 0.567 0 1 1
1973 34 68 328690 41434 0.126 5.12 7.63 0.681 0 1 1

Appendix B
Telegraph and telephone 1954 24 77 94159 19913 0.211 1.45 1.66 0.509 0 2 0
apparatus and equipment 1958 26 86 109117 17372 0.159 1.85 2.09 0.499 0 2 0
(MLH 363) 1963 36 86 153897 33656 0.219 2.85 2.78 0.528 0 2 0
Radio and other electronic 1954 236 407 227325 40139 0.177 0.54 1.69 0.509 0 2 0

Data Sets
equipment, except valves and 1958 250 432 297348 45764 0.154 0.62 1.80 0.499 0 2 0
semiconductors (1958 S.I.C.) 1963 271 546 492174 128292 0.261 0.65 1.75 0.528 0 2 0
Radio and electronic 1963 119 179 73362 18597 0.253 0.39 1.45 0.528 0 1 0
components, except valves and 1968 167 256 179345 52172 0.291 0.53 1.85 0.567 0 1 0
semiconductors (1968 S.I.C.)
Broadcast receiving and sound 1963 50 83 131206 22732 0.173 0.48 1.07 0.528 1 1 0
reproducing equipment, except 1968 33 66 109384 19626 0.179 0.73 1.76 0.567 1 1 0
records and tapes
Electronic computers (MLH 366) 1963 11 27 39154 12371 0.316 1.80 4.78 0.528 0 2 0
1968 20 36 92590 21326 0.230 2.24 4.33 0.567 0 2 0
1973 35 46 208460 63662 0.305 2.62 4.72 0.681 0 2 0
Radio, radar, and electronic 1963 77 168 171326 46895 0.274 1.25 2.87 0.528 0 2 0
capital goods (MLH 367) 1968 94 195 212221 49073 0.231 1.34 3.42 0.567 0 2 0
1973 115 201 289685 70163 0.242 1.61 3.75 0.681 0 2 0
Electric appliances primarily for 1954 92 140 95030 22944 0.241 0.68 2.24 0.509 2 1 1
domestic use (MLH 368) 1958 85 115 121328 23091 0.190 1.12 2.60 0.499 2 1 1
1963 102 142 211018 49486 0.235 1.16 2.32 0.528 2 1 1
1968 100 153 240736 56430 0.234 1.48 3.07 0.567 2 1 1
1973 92 119 276881 57904 0.209 2.25 4.16 0.681 2 1 1
Electrical equipment for motor 1954 33 50 63222 14327 0.227 1.11 1.96 0.509 0 1 0
vehicles, cycles, and aircraft, 1958 29 49 70874 11050 0.156 1.34 2.10 0.499 0 1 0
excluding accumulators 1963 33 63 74217 17484 0.236 1.17 2.26 0.528 0 1 0
1968 31 74 128479 35546 0.277 1.44 2.31 0.567 0 1 0
1973 31 — 138784 22775 0.164 — 2.87 0.681 0 1 0

487
Table B6 (continued)

488
Industry Year NFIRMS NPLANTS SS PROFIT PCM K/N K/L UNION ADTYPE RDTYPE CHANGE
Secondary batteries 1954 14 23 24832 4827 0.194 0.50 1.46 0.509 1 0 1
(accumulators) 1958 16 21 23954 4507 0.188 0.67 1.68 0.499 1 0 1
1963 13 17 27780 5916 0.213 0.94 1.86 0.528 1 0 1
1968 18 26 40769 9680 0.237 1.28 3.03 0.567 1 0 1
Electric light fittings, wiring 1954 153 187 39233 8962 0.228 0.20 1.48 0.509 0 0 1
accessories, and other electrical 1958 168 195 59906 12937 0.216 0.23 1.38 0.499 0 0 1
goods
1963 160 192 73950 18626 0.252 0.30 1.79 0.528 0 0 1
1968 129 163 80961 20199 0.249 0.42 2.29 0.567 0 0 1
Shipbuilding and marine 1954 125 202 360296 35662 0.099 1.17 1.40 0.509 0 0 0
engineering 1958 148 216 384318 37087 0.097 1.23 1.56 0.499 0 0 0
1963 102 161 285071 32624 0.114 1.71 2.42 0.528 0 0 0
1968 81 146 223519 26045 0.117 1.56 2.26 0.567 0 0 0
Boatbuilding 1954 59 68 8992 1398 0.155 0.04 0.45 0.509 0 0 0
1958 54 62 6122 681 0.111 0.03 0.43 0.499 0 0 0
1963 46 55 7248 1079 0.149 0.02 0.33 0.528 0 0 0
1968 51 56 11329 2130 0.188 0.06 0.92 0.567 0 0 0
Cars 1954 11 25 401716 55719 0.139 27.41 9.73 0.509 1 1 0
1958 12 29 662249 75580 0.114 29.64 8.93 0.499 1 1 0
1963 15 52 1042379 170884 0.164 23.04 7.98 0.528 1 1 0

Appendix B
1968 11 43 1107976 141402 0.128 37.04 9.47 0.567 1 1 0
Motor bodies 1954 106 145 132935 15479 0.116 1.72 4.27 0.509 0 0 0

Data Sets
1958 100 131 194393 17912 0.092 2.43 4.92 0.499 0 0 0
1963 82 117 180054 22467 0.125 3.12 5.95 0.528 0 0 0
1968 87 124 150358 20159 0.134 2.08 5.09 0.567 0 0 0
Internal combustion engines for 1954 6 8 36161 5869 0.162 7.57 5.01 0.509 0 1 0
vehicles 1958 7 8 41374 3685 0.089 6.91 3.75 0.499 0 1 0
1963 6 10 59125 9306 0.157 5.47 3.32 0.528 0 1 0
Parts and accessories of motor 1954 169 219 158319 29916 0.189 2.06 7.01 0.509 0 0 0
vehicles, other than engines and 1958 142 187 182819 32714 0.179 2.18 6.25 0.499 0 0 0
motor bodies
1963 142 239 317941 62425 0.196 1.87 4.55 0.528 0 0 0
Hand tools and implements 1954 150 175 27296 5265 0.193 0.26 2.38 0.509 0 0 1
(MLH 391) 1958 121 157 30019 6930 0.231 0.29 2.46 0.499 0 0 1
1963 105 138 33132 8260 0.249 0.35 2.93 0.528 0 0 1
1968 94 122 33539 7879 0.235 0.42 3.26 0.567 0 0 1
1973 86 103 38787 10186 0.263 0.57 4.15 0.681 0 0 1
Razors and blades 1954 4 8 11773 7277 0.618 1.59 3.85 0.509 2 1 0
1958 5 7 11825 6818 0.577 2.33 5.14 0.499 2 1 0
1963 4 5 16437 8174 0.497 6.66 6.13 0.528 2 1 0
1968 4 5 20533 12372 0.603 7.66 10.17 0.567 2 1 0
Wire and wire manufactures 1954 169 231 133773 15781 0.118 0.53 3.70 0.509 0 0 1
(MLH 394) 1958 155 225 165482 18115 0.109 0.60 3.80 0.499 0 0 1
1963 141 246 184890 24667 0.133 0.72 4.51 0.528 0 0 1
1968 125 213 186584 24964 0.134 0.98 5.77 0.567 0 0 1
1973 120 193 236630 42727 0.181 1.25 6.49 0.681 0 0 1

489
Table B6 (continued)

490
Industry Year NFIRMS NPLANTS SS PROFIT PCM K/N K/L UNION ADTYPE RDTYPE CHANGE
Cans and metal boxes (MLH 395) 1954 69 105 70094 8992 0.128 0.67 2.48 0.509 0 0 0
1958 63 96 82301 12013 0.146 0.89 3.12 0.499 0 0 0
1963 44 78 101514 12318 0.121 1.58 4.16 0.528 0 0 0
1968 44 75 116529 16418 0.141 2.36 6.22 0.567 0 0 0
1973 44 70 138612 15881 0.115 3.05 7.39 0.681 0 0 0
Jewellery and plate (1958 S.I.C.) 1954 199 225 25515 5400 0.212 0.08 1.15 0.509 0 0 0
1958 153 174 23848 4969 0.208 0.11 1.31 0.499 0 0 0
1963 139 166 34007 6943 0.204 0.13 1.52 0.528 0 0 0
Jewellery and plate (1968 S.I.C.) 1963 101 118 24399 4602 0.189 0.14 1.94 0.528 0 0 0
1968 67 75 29107 4800 0.165 0.38 4.45 0.567 0 0 0
Metal furniture 1954 123 142 33109 6239 0.188 0.21 1.84 0.509 0 0 0
1958 100 116 31832 5703 0.179 0.29 2.54 0.499 0 0 0
1963 118 146 50613 12094 0.239 0.27 2.01 0.528 0 0 0
1968 107 138 58593 11913 0.203 0.35 2.47 0.567 0 0 0
1973 109 133 77913 16159 0.207 0.43 2.84 0.681 0 0 0
Drop forgings of steel and iron, 1963 87 106 88475 14464 0.163 1.38 5.37 0.528 0 0 1
and steel stampings and 1968 62 91 89201 14181 0.159 1.51 5.38 0.567 0 0 1
pressings
1973 65 103 105412 17705 0.168 1.87 7.17 0.681 0 0 1
Domestic hollow-ware 1954 101 117 29105 5797 0.199 0.21 1.50 0.509 0 0 1

Appendix B
1958 84 93 19498 2363 0.121 0.25 2.03 0.499 0 0 1
1963 72 84 22866 3964 0.173 0.25 1.88 0.528 0 0 1
1968 50 61 19697 4876 0.248 0.43 3.57 0.567 0 0 1
Industrial hollow-ware 1954 81 93 27689 4795 0.173 0.32 2.94 0.509 0 0 1

Data Sets
1958 94 118 44388 6654 0.150 0.39 2.77 0.499 0 0 1
1963 60 78 32322 4926 0.152 0.66 5.31 0.528 0 0 1
1968 45 65 31439 5845 0.186 0.75 5.80 0.567 0 0 1
Metal windows and door frames 1954 34 46 26320 3254 0.124 0.46 1.64 0.509 0 0 1
1958 35 44 32013 5152 0.161 0.48 1.43 0.499 0 0 1
1963 39 56 33880 6678 0.197 0.41 1.56 0.528 0 0 1
1968 39 60 35126 8096 0.230 0.36 1.66 0.567 0 0 1
Safes, locks, latches, and keys 1963 39 45 15359 5154 0.336 0.35 1.87 0.528 0 0 1
1968 32 42 19913 5848 0.294 0.66 2.84 0.567 0 0 1
Engineers’ and mechanicians’ 1963 56 82 32464 7387 0.228 0.73 4.61 0.528 0 0 1
goods 1968 46 61 43906 9840 0.224 1.08 5.32 0.567 0 0 1
Anchors and chains 1954 37 50 9541 1729 0.181 0.24 2.69 0.509 0 0 0
1958 30 41 8906 1244 0.140 0.40 4.40 0.499 0 0 0
1963 27 33 10021 1811 0.181 0.68 7.07 0.528 0 0 0
1968 19 25 9103 1746 0.192 0.71 7.63 0.567 0 0 0
Spun cotton yarns 1954 136 287 190888 20326 0.106 0.65 2.35 0.752 0 0 1
1958 102 231 104674 9038 0.086 0.85 3.27 0.720 0 0 1
1963 51 111 64778 6175 0.095 1.80 6.03 0.790 0 0 1
1968 40 77 52826 4553 0.086 2.12 7.92 0.810 0 0 1
Cotton waste yarns 1954 48 68 16087 2799 0.174 0.23 2.87 0.752 0 0 1
1958 45 61 11765 2075 0.176 0.25 2.94 0.720 0 0 1
1963 31 44 10170 1275 0.125 0.33 3.02 0.790 0 0 1
1968 19 25 4273 480 0.112 0.40 4.60 0.810 0 0 1

491
Table B6 (continued)

492
Industry Year NFIRMS NPLANTS SS PROFIT PCM K/N K/L UNION ADTYPE RDTYPE CHANGE
Spun man-made fibres and 1954 32 40 21005 3105 0.148 0.63 2.54 0.752 0 0 1
mixture yarns 1958 32 42 20354 1263 0.062 0.59 2.32 0.720 0 0 1
1963 24 37 20716 2470 0.119 1.02 4.11 0.790 0 0 1
1968 18 36 33794 4877 0.144 1.59 5.15 0.810 0 0 1
Creped, bulked, textured, or 1963 26 38 26982 3552 0.132 2.11 11.37 0.790 0 0 0
stretch continuous filament yarns 1968 27 53 66673 14572 0.219 4.00 13.94 0.810 0 0 0
Finished thread 1954 31 45 30305 4211 0.139 0.65 1.97 0.752 0 0 0
1958 24 38 24631 3851 0.156 1.19 4.07 0.720 0 0 0
1963 24 37 23646 3847 0.163 2.18 7.97 0.790 0 0 0
1968 18 28 29301 9796 0.334 2.77 8.86 0.810 0 0 0
Loom state cloth of cotton 1954 402 556 174219 13237 0.076 0.28 2.19 0.752 0 0 0
1958 252 362 97130 6347 0.065 0.44 3.30 0.720 0 0 0
1963 119 164 54511 4164 0.076 0.61 4.59 0.790 0 0 0
1968 73 99 34264 2872 0.084 0.70 5.64 0.810 0 0 0
Loom state cloth of man-made 1954 162 209 77785 7233 0.093 0.52 3.20 0.752 0 0 1
fibres and mixtures 1958 110 149 56182 4269 0.076 0.62 4.12 0.720 0 0 1
1963 69 102 62504 4982 0.080 1.24 6.86 0.790 0 0 1
1968 52 78 52929 5997 0.113 2.33 14.58 0.810 0 0 1
Wool combers (on own account) 1954 10 12 31294 1867 0.060 0.45 1.82 0.320 0 0 0

Appendix B
1958 14 16 28775 2226 0.077 0.64 3.02 0.314 0 0 0
1963 18 35 47279 4627 0.098 0.70 4.09 0.310 0 0 0
Wool combers (on commission) 1954 39 59 13443 4497 0.335 0.49 3.00 0.320 0 0 1

Data Sets
1958 34 55 12858 3567 0.277 0.80 4.63 0.314 0 0 1
1963 26 47 14971 4647 0.310 0.93 5.15 0.310 0 0 1
Spun woollen yarns 1954 91 103 44201 4728 0.107 0.34 3.78 0.320 0 0 0
1958 85 103 38705 5130 0.133 0.34 3.65 0.314 0 0 0
1963 83 108 61631 11066 0.180 0.46 3.46 0.310 0 0 0
1968 80 108 60494 11362 0.188 0.81 6.05 0.346 0 0 0
Woven worsted fabric 1954 222 301 123535 16119 0.130 0.26 1.96 0.320 0 0 0
1958 182 252 96368 10333 0.107 0.28 2.13 0.314 0 0 0
1963 130 229 95805 13226 0.138 0.34 2.23 0.310 0 0 0
1968 83 171 73266 12630 0.172 0.49 3.49 0.346 0 0 0
Woven woollen fabric, other 1954 253 324 116771 18863 0.162 0.34 2.08 0.320 0 0 0
than blankets, traveling rugs, etc. 1958 217 308 95201 11959 0.126 0.31 1.97 0.314 0 0 0
1963 162 237 78752 15308 0.194 0.36 2.27 0.310 0 0 0
1968 135 200 73924 14532 0.197 0.52 3.23 0.346 0 0 0
Mechanical cloth, woven felt, 1954 20 33 15088 1527 0.101 0.54 3.78 0.320 0 0 1
and pressed felt 1958 16 27 11508 2611 0.227 0.59 3.90 0.314 0 0 1
1963 13 20 8972 2232 0.249 0.74 4.88 0.310 0 0 1
1968 13 22 11636 3318 0.285 0.97 5.95 0.346 0 0 1
Jute (MLH 415) 1954 47 72 43789 6016 0.137 0.86 3.20 0.320 0 0 1
1958 48 74 38026 3252 0.086 0.91 4.14 0.314 0 0 1
1963 32 64 38058 5190 0.136 1.02 4.03 0.310 0 0 1
1968 28 57 39104 6035 0.154 1.24 5.03 0.346 0 0 1
1973 21 40 39220 6997 0.178 1.91 7.67 0.449 0 0 1

493
Table B6 (continued)

494
Industry Year NFIRMS NPLANTS SS PROFIT PCM K/N K/L UNION ADTYPE RDTYPE CHANGE
Rope, twine, and net (MLH 416) 1954 82 127 32957 4595 0.139 0.21 1.94 0.320 0 0 1
1958 64 110 27823 4633 0.167 0.28 2.55 0.314 0 0 1
1963 47 99 31223 5564 0.178 0.34 3.14 0.310 0 0 1
1968 34 54 23690 4720 0.199 0.70 4.73 0.346 0 0 1
1973 23 40 22809 4383 0.192 1.09 7.12 0.449 0 0 1
Hosiery 1954 240 321 105983 22583 0.213 0.94 5.85 0.320 1 0 0
1958 188 248 79577 13304 0.167 1.14 6.98 0.314 1 0 0
1963 157 202 65001 13436 0.207 1.04 6.92 0.310 1 0 0
1968 92 148 78294 17925 0.229 1.06 5.36 0.346 1 0 0
Underwear, shirts, and 1954 100 144 38471 6014 0.156 0.23 1.59 0.320 0 0 0
nightwear, knitted, netted, or 1958 90 139 38043 6165 0.162 0.33 2.20 0.314 0 0 0
crocheted 1963 58 114 38649 6750 0.175 0.49 2.81 0.310 0 0 0
1968 52 101 50444 12221 0.242 0.60 3.00 0.346 0 0 0
Other garments, etc., knitted, 1954 326 408 62220 11085 0.178 0.15 1.65 0.320 0 0 0
netted, or crocheted 1958 293 379 70102 11562 0.165 0.28 2.54 0.314 0 0 0
1963 257 394 114945 21303 0.185 0.49 3.31 0.310 0 0 0
1968 245 410 138644 27366 0.197 0.64 4.17 0.346 0 0 0
Woven carpets, carpeting, and 1954 52 84 76253 14337 0.188 0.99 2.94 0.320 0 0 1
carpet floor rugs 1958 54 91 78295 14358 0.183 0.94 2.98 0.314 0 0 1

Appendix B
1963 42 68 79534 16127 0.203 1.24 3.22 0.310 0 0 1
1968 32 68 80275 15588 0.194 1.25 3.67 0.346 0 0 1
Tufted carpets, carpeting, and 1958 5 7 4748 618 0.130 0.90 5.98 0.314 1 0 0

Data Sets
carpet floor rugs 1963 20 24 21159 3938 0.186 0.85 5.20 0.310 1 0 0
1968 27 37 56536 6529 0.115 1.44 5.41 0.346 1 0 0
Made-up household textiles and 1954 215 244 34227 3645 0.106 0.04 0.49 0.320 0 0 0
handkerchiefs (MLH 422.1) 1958 204 254 35580 4518 0.127 0.05 0.65 0.314 0 0 0
1963 190 232 41617 5673 0.136 0.07 0.89 0.310 0 0 0
1968 147 186 46617 6591 0.141 0.14 1.60 0.346 0 0 0
1973 120 194 69003 14772 0.214 0.23 2.34 0.449 0 0 0
Canvas sacks and bags 1954 80 115 30866 1582 0.051 0.05 0.90 0.320 0 0 1
1958 68 95 20322 2217 0.109 0.06 1.15 0.314 0 0 1
1963 47 71 18744 1247 0.067 0.06 1.11 0.310 0 0 1
1968 31 39 7558 777 0.103 0.08 1.64 0.346 0 0 1
Textile finishing: yarn 1954 92 111 13542 2238 0.165 0.27 2.93 0.320 0 0 1
1958 70 87 10760 1835 0.171 0.32 3.67 0.314 0 0 1
1963 52 68 10634 2594 0.244 0.50 4.86 0.310 0 0 1
1968 41 56 10861 2770 0.255 0.86 8.49 0.346 0 0 1
Textile finishing: woven fabrics 1954 179 241 64563 13743 0.213 0.47 2.61 0.320 0 0 1
of cotton and man-made fibres 1958 151 220 58328 8681 0.149 0.65 3.58 0.314 0 0 1
1963 96 167 48969 10003 0.204 0.90 5.04 0.310 0 0 1
1968 76 123 42515 10183 0.240 1.28 7.39 0.346 0 0 1
Textile finishing: woollen and 1954 60 71 9624 2534 0.263 0.36 4.44 0.320 0 0 1
worsted fabrics 1958 45 52 6744 1836 0.272 0.38 5.50 0.314 0 0 1
1963 37 45 6617 1658 0.251 0.30 3.64 0.310 0 0 1
1968 23 29 4783 1282 0.268 0.49 5.84 0.346 0 0 1

495
Table B6 (continued)

496
Industry Year NFIRMS NPLANTS SS PROFIT PCM K/N K/L UNION ADTYPE RDTYPE CHANGE
Textile finishing: knitted fabrics 1954 64 75 10559 2907 0.275 0.36 2.91 0.320 0 0 1
and other knitted goods 1958 51 63 10904 2854 0.262 0.52 3.74 0.314 0 0 1
1963 44 57 15408 4580 0.297 0.87 4.94 0.310 0 0 1
1968 50 70 23572 7396 0.314 1.08 5.93 0.346 0 0 1
Asbestos manufactures 1954 29 40 42462 9203 0.217 1.18 2.95 0.320 0 0 1
(MLH 429.1) 1958 30 41 45229 11116 0.246 1.59 3.60 0.314 0 0 1
1963 23 40 55278 13314 0.241 1.80 3.60 0.310 0 0 1
1968 20 39 66278 15944 0.241 2.24 4.29 0.346 0 0 1
1973 16 33 65727 14066 0.214 3.01 5.55 0.449 0 0 1
Needle felt and needleloom 1954 13 17 5514 1168 0.212 0.14 2.20 0.320 0 0 0
carpeting 1958 13 21 4734 713 0.151 0.14 2.50 0.314 0 0 0
1963 15 25 7124 1277 0.179 0.19 2.94 0.310 0 0 0
1968 14 24 8298 2177 0.262 0.40 5.02 0.346 0 0 0
Leather tanning 1954 72 85 38382 2754 0.072 0.15 1.50 0.330 0 0 0
1958 49 56 21405 1666 0.078 0.20 2.05 0.323 0 0 0
1963 29 32 17045 1938 0.114 0.31 2.63 0.318 0 0 0
1968 22 25 15512 1783 0.115 0.50 4.13 0.284 0 0 0
Leather tanning and dressing, or 1954 170 194 65312 6249 0.096 0.22 2.33 0.330 0 0 0
dressing only 1958 150 172 56692 5827 0.103 0.23 2.43 0.323 0 0 0

Appendix B
1963 130 156 59245 8234 0.139 0.27 2.82 0.318 0 0 0
1968 120 140 67775 11078 0.163 0.36 3.63 0.284 0 0 0
Fellmongery 1954 25 27 11148 1103 0.099 0.13 2.48 0.330 0 0 0

Data Sets
1958 22 24 8095 549 0.068 0.16 2.90 0.323 0 0 0
1963 29 37 17194 1588 0.092 0.14 2.68 0.318 0 0 0
1968 20 24 11301 1117 0.099 0.22 3.89 0.284 0 0 0
Travel goods of leather or leather 1954 50 56 8974 1632 0.182 0.06 0.61 0.330 0 0 0
substitutes 1958 33 38 6808 1247 0.183 0.08 0.81 0.323 0 0 0
1963 31 36 7735 1557 0.201 0.11 0.99 0.318 0 0 0
1968 38 45 10300 1823 0.177 0.11 0.99 0.284 0 0 0
Leather goods other than travel 1954 157 176 13442 2024 0.151 0.02 0.36 0.330 0 0 0
goods 1958 125 134 11890 2042 0.172 0.03 0.49 0.323 0 0 0
1963 120 131 16553 2965 0.179 0.05 0.75 0.318 0 0 0
1968 110 130 16823 3272 0.194 0.06 0.92 0.284 0 0 0
Fur (MLH 433) 1954 72 84 9069 1554 0.171 0.08 1.39 0.330 0 0 0
1958 52 55 8614 1158 0.134 0.11 1.62 0.323 0 0 0
1963 53 65 15239 3253 0.213 0.11 1.49 0.318 0 0 0
1968 46 55 13112 2123 0.162 0.14 2.14 0.284 0 0 0
1973 41 57 11104 2121 0.191 0.15 2.69 0.267 0 0 0
Weatherproof outerwear (MLH 1954 — 304 45817 5898 0.129 0.07 0.69 0.337 0 0 0
441) 1958 180 242 35686 4591 0.129 0.08 0.83 0.326 0 0 0
1963 157 232 44065 7353 0.167 0.09 0.82 0.319 0 0 0
1968 118 186 35433 7102 0.200 0.10 1.06 0.333 0 0 0
1973 120 178 34884 6999 0.201 0.12 1.29 0.369 0 0 0

497
Table B6 (continued)

498
Industry Year NFIRMS NPLANTS SS PROFIT PCM K/N K/L UNION ADTYPE RDTYPE CHANGE
Men’s and boys’ tailored 1954 544 760 168227 21008 0.125 0.05 0.33 0.337 0 0 0
outerwear (MLH 442) 1958 459 652 153941 17986 0.117 0.07 0.39 0.326 0 0 0
1963 352 592 163246 23884 0.146 0.09 0.50 0.319 0 0 0
1968 302 551 161643 27406 0.170 0.12 0.72 0.333 0 0 0
1973 315 513 180010 34679 0.193 0.16 0.93 0.369 0 0 0
Women’s and girls’ tailored 1954 521 635 86800 11507 0.133 0.05 0.60 0.337 0 0 0
outerwear (MLH 443) 1958 406 520 74984 9193 0.123 0.06 0.66 0.326 0 0 0
1963 369 465 83662 12967 0.155 0.07 0.75 0.319 0 0 0
1968 279 382 80623 13760 0.171 0.10 1.02 0.333 0 0 0
1973 355 416 87260 18909 0.217 0.10 1.15 0.369 0 0 0
Heavy overalls and jeans 1954 95 127 17596 1751 0.100 0.05 0.59 0.337 0 0 0
1958 76 99 16160 1871 0.116 0.07 0.65 0.326 0 0 0
1963 81 114 20516 3469 0.169 0.06 0.66 0.319 0 0 0
1968 73 112 21566 3769 0.175 0.08 0.76 0.333 0 0 0
Shirts, underwear, etc. (on own 1954 213 273 44879 5099 0.114 0.05 0.44 0.337 1 0 0
account) 1958 181 246 45719 5243 0.115 0.06 0.51 0.326 1 0 0
1963 149 227 52713 7485 0.142 0.09 0.72 0.319 1 0 0
1968 111 183 51955 8750 0.168 0.15 1.06 0.333 1 0 0
Women’s and girls’ light 1954 549 665 68151 8938 0.131 0.03 0.43 0.337 0 0 0

Appendix B
outerwear (on own account) 1958 492 611 69599 9010 0.129 0.03 0.46 0.326 0 0 0
1963 329 453 63530 10908 0.172 0.05 0.65 0.319 0 0 0
1968 272 403 86700 14397 0.166 0.07 0.74 0.333 0 0 0
Women’s and girls’ underwear 1954 148 190 24229 2692 0.111 0.05 0.57 0.337 0 0 0

Data Sets
and nightwear 1958 146 186 30385 2642 0.087 0.06 0.62 0.326 0 0 0
1963 122 172 33041 4452 0.135 0.10 0.86 0.319 0 0 0
1968 118 186 33513 6074 0.181 0.13 1.36 0.333 0 0 0
Infants’ wear 1954 127 149 12564 2029 0.161 0.03 0.41 0.337 0 0 0
1958 119 152 14414 2316 0.161 0.03 0.49 0.326 0 0 0
1963 94 132 13504 2195 0.163 0.06 0.78 0.319 0 0 0
1968 93 140 18789 3629 0.193 0.08 1.04 0.333 0 0 0
Corsetry 1954 69 105 21817 4559 0.209 0.20 1.40 0.337 2 0 0
1958 62 101 28208 6340 0.225 0.22 1.27 0.326 2 0 0
1963 63 123 36944 8763 0.237 0.19 1.19 0.319 2 0 0
1968 53 119 40566 10123 0.250 0.20 1.25 0.333 2 0 0
Gloves (MLH 449.2) 1954 120 188 14406 2688 0.187 0.03 0.59 0.337 0 0 0
1958 89 139 10505 1798 0.171 0.04 0.80 0.326 0 0 0
1963 76 122 12731 2387 0.187 0.05 0.84 0.319 0 0 0
1968 55 92 11338 2077 0.183 0.07 1.11 0.333 0 0 0
1973 55 80 12770 2627 0.206 0.09 1.45 0.369 0 0 0
Footwear (MLH 450) 1954 529 778 192337 28566 0.149 0.09 0.63 0.718 1 0 0
1958 457 674 176126 25157 0.143 0.11 0.72 0.724 1 0 0
1963 370 615 206016 37473 0.182 0.15 0.91 0.731 1 0 0
1968 299 559 207150 38775 0.187 0.20 1.21 0.779 1 0 0
1973 266 441 238701 47744 0.200 0.29 1.54 0.813 1 0 0
Refractory goods 1954 88 147 33677 5805 0.172 0.25 2.25 0.268 0 0 1
1958 73 149 37460 6518 0.174 0.42 3.95 0.257 0 0 1
1963 63 130 40633 8734 0.215 0.63 6.02 0.256 0 0 1

499
1968 44 111 52710 12125 0.230 0.71 5.67 0.289 0 0 1
1973 47 — 60518 16273 0.269 — 6.56 0.385 0 0 1
Table B6 (continued)

500
Industry Year NFIRMS NPLANTS SS PROFIT PCM K/N K/L UNION ADTYPE RDTYPE CHANGE
Building bricks 1954 309 543 60779 12568 0.207 0.12 1.70 0.268 0 0 1
1958 224 525 54900 9440 0.172 0.12 1.87 0.257 0 0 1
1963 179 433 71157 16717 0.235 0.20 2.49 0.256 0 0 1
1968 109 324 64763 15739 0.243 0.45 5.25 0.289 0 0 1
Sanitary ware of fireclay, etc. 1954 82 97 17909 3339 0.186 0.14 1.15 0.268 0 0 1
1958 70 85 13784 2066 0.150 0.19 1.75 0.257 0 0 1
1963 60 81 17383 3606 0.207 0.26 2.34 0.256 0 0 1
1968 35 72 17796 3823 0.215 0.42 3.88 0.289 0 0 1
Tiles, pipes, and other products 1954 62 84 9536 1714 0.180 0.17 1.91 0.268 0 0 1
of fireclay, etc. 1958 52 96 9312 1384 0.149 0.13 1.96 0.257 0 0 1
1963 32 51 9728 2302 0.237 0.24 2.53 0.256 0 0 1
1968 28 37 11420 2728 0.239 0.45 3.79 0.289 0 0 1
Tiles, other than of precast 1954 36 44 12799 2808 0.219 0.54 2.36 0.574 0 0 0
concrete and brick earth 1958 25 32 11454 2234 0.195 0.79 3.17 0.524 0 0 0
1963 17 28 16758 3863 0.231 1.52 5.04 0.569 0 0 0
1968 10 22 19210 3499 0.182 2.78 8.09 0.721 0 0 0
Sanitary earthenware 1954 16 21 7719 2117 0.274 0.28 1.33 0.574 0 0 1
1958 14 17 6663 1502 0.225 0.59 2.34 0.524 0 0 1
1963 14 18 11562 2818 0.244 0.80 2.46 0.569 0 0 1

Appendix B
1968 7 14 12762 3348 0.262 0.74 2.20 0.721 0 0 1
China, earthenware, etc., 1954 117 143 30243 6367 0.211 0.29 1.11 0.574 0 0 1

Data Sets
domestic and ornamental 1958 116 152 32197 6611 0.205 0.26 1.05 0.524 0 0 1
1963 94 132 35215 7119 0.202 0.26 0.95 0.569 0 0 1
1968 76 121 40676 9435 0.232 0.32 1.15 0.721 0 0 1
Electrical ware of porcelain, 1954 14 20 6562 1456 0.222 0.58 1.95 0.574 0 0 1
earthenware, or stoneware 1958 12 18 8122 2027 0.250 0.83 2.53 0.524 0 0 1
1963 13 22 10133 2653 0.262 1.15 3.58 0.569 0 0 1
1968 10 20 10413 1995 0.192 1.23 3.85 0.721 0 0 1
Glass containers 1954 38 58 41369 7882 0.191 1.22 2.98 0.411 0 0 1
1958 32 51 48845 9591 0.196 1.69 3.49 0.394 0 0 1
1963 23 41 56727 10781 0.190 2.27 3.91 0.422 0 0 1
1968 19 37 62587 13997 0.224 3.02 5.15 0.524 0 0 1
Domestic and fancy glassware 1954 19 26 7062 1858 0.263 0.11 0.47 0.411 1 0 0
1958 16 20 7544 1563 0.207 0.29 1.00 0.394 1 0 0
1963 15 18 9446 3088 0.327 0.41 1.32 0.422 1 0 0
1968 16 24 18584 5410 0.291 0.67 1.67 0.524 1 0 0
Glass fibre and miscellaneous 1954 15 22 6283 1995 0.318 0.55 4.49 0.411 0 1 0
glass products 1958 22 34 10631 3104 0.292 0.43 3.40 0.394 0 1 0
1963 29 40 14873 4490 0.302 0.66 5.10 0.422 0 1 0
Glass fibre and manufactures 1963 9 12 9683 3334 0.344 2.45 9.56 0.422 0 2 0
thereof 1968 7 10 12907 5270 0.408 3.22 9.15 0.524 0 2 0
Cement (MLH 464) 1954 13 47 68489 19738 0.288 2.52 9.41 0.268 0 0 0
1958 13 50 70011 17720 0.253 2.71 10.64 0.257 0 0 0
1963 14 54 83280 26467 0.318 3.79 14.55 0.256 0 0 0
1968 9 53 89608 25912 0.289 6.26 24.08 0.289 0 0 0

501
1973 8 53 127577 39435 0.309 7.45 27.07 0.385 0 0 0
Table B6 (continued)

502
Industry Year NFIRMS NPLANTS SS PROFIT PCM K/N K/L UNION ADTYPE RDTYPE CHANGE
Lime and whiting 1954 37 45 9671 2038 0.211 0.60 6.71 0.268 0 0 1
1958 33 49 9989 2392 0.239 0.63 8.74 0.257 0 0 1
1963 26 35 9999 2776 0.278 0.71 8.92 0.256 0 0 1
1968 11 17 11626 2421 0.208 2.53 20.61 0.289 0 0 1
Pre-cast concrete goods 1954 258 362 48767 8572 0.176 0.14 2.01 0.268 0 0 1
1958 206 312 49582 8490 0.171 0.19 2.63 0.257 0 0 1
1963 201 337 65103 13518 0.208 0.23 3.06 0.256 0 0 1
1968 229 384 107266 25685 0.239 0.41 5.04 0.289 0 0 1
Asbestos cement goods 1954 8 20 21662 4755 0.220 1.18 3.11 0.268 0 0 1
1958 7 21 20931 4575 0.219 1.25 3.61 0.257 0 0 1
1963 8 23 26577 6613 0.249 1.47 3.94 0.256 0 0 1
1968 8 22 27503 7212 0.262 2.06 5.53 0.289 0 0 1
Builders’ woodwork and 1954 180 202 39392 4570 0.116 0.09 0.89 0.456 0 0 1
prefabricated timber buildings 1958 147 173 40900 5089 0.124 0.12 1.11 0.435 0 0 1
1963 173 208 72485 14250 0.197 0.16 1.38 0.394 0 0 1
1968 228 282 103423 19090 0.185 0.18 1.70 0.323 0 0 1
Wood chipboard 1963 9 16 6603 1387 0.210 0.78 6.17 0.394 0 0 0
1968 11 18 11087 2166 0.195 1.04 6.39 0.323 0 0 0
Plywood and textured boards 1963 17 17 7218 569 0.079 0.41 3.17 0.394 0 0 0

Appendix B
1968 19 24 17348 2560 0.148 0.43 3.41 0.323 0 0 0
Domestic furniture, not 1954 340 424 90625 12637 0.139 0.11 0.98 0.456 0 0 0

Data Sets
upholstered, mainly of wood 1958 265 389 81017 11896 0.147 0.12 1.06 0.435 0 0 0
1963 238 380 89233 15734 0.176 0.13 1.16 0.394 0 0 0
1968 237 298 111984 25856 0.231 0.24 1.98 0.323 0 0 0
Upholstered furniture 1954 207 249 41073 6001 0.146 0.03 0.43 0.456 0 0 0
1958 155 190 41476 5678 0.137 0.05 0.54 0.435 0 0 0
1963 159 195 55426 9830 0.177 0.07 0.70 0.394 0 0 0
1968 162 200 56159 10789 0.192 0.10 1.13 0.323 0 0 0
Office, school, and other 1954 174 199 28139 4339 0.154 0.08 0.91 0.456 0 0 0
furniture 1958 179 216 36409 5271 0.145 0.10 1.04 0.435 0 0 0
1963 137 162 33916 5949 0.175 0.17 1.73 0.394 0 0 0
1968 112 132 36408 6781 0.186 0.20 1.93 0.323 0 0 0
Bedding 1954 108 145 28579 4078 0.143 0.15 1.94 0.456 1 0 0
1958 72 103 26141 3686 0.141 0.19 2.13 0.435 1 0 0
1963 55 53 30032 5471 0.182 0.35 2.06 0.394 1 0 0
1968 53 90 44569 8526 0.191 0.20 1.63 0.323 1 0 0
Wooden containers and baskets 1954 271 332 39675 4607 0.116 0.05 0.87 0.456 0 0 0
(MLH 475) 1958 192 229 29239 3828 0.131 0.08 1.34 0.435 0 0 0
1963 165 216 30027 4554 0.152 0.09 1.55 0.394 0 0 0
1968 147 197 31349 5388 0.172 0.11 1.92 0.323 0 0 0
1973 160 228 49112 10133 0.206 0.12 2.27 0.340 0 0 0
Cork manufactures 1954 19 25 3022 575 0.190 0.10 1.71 0.456 0 0 0
1958 13 18 3348 576 0.172 0.25 3.39 0.435 0 0 0
1963 8 9 3855 1013 0.263 0.59 3.87 0.394 0 0 0
1968 7 10 4116 812 0.197 0.47 3.98 0.323 0 0 0

503
Table B6 (continued)

504
Industry Year NFIRMS NPLANTS SS PROFIT PCM K/N K/L UNION ADTYPE RDTYPE CHANGE
Paper and board (MLH 481) 1954 186 253 311039 71029 0.228 1.83 6.37 0.367 0 0 1
1958 167 262 332236 57771 0.174 2.74 8.81 0.361 0 0 1
1963 144 267 381309 74910 0.196 3.18 9.66 0.377 0 0 1
1968 111 207 385640 71760 0.186 5.05 13.98 0.403 0 0 1
1973 92 171 415335 74153 0.179 6.31 18.04 0.481 0 0 1
Rigid boxes 1954 211 252 25264 4885 0.193 0.12 1.54 0.367 0 0 0
1958 177 216 21835 4041 0.185 0.13 1.65 0.361 0 0 0
1963 154 188 21606 4474 0.207 0.12 1.50 0.377 0 0 0
1968 117 147 19616 3891 0.198 0.16 2.05 0.403 0 0 0
Cartons 1954 71 92 42532 7823 0.184 0.79 3.68 0.367 0 0 0
1958 72 91 52073 7774 0.149 1.07 4.35 0.361 0 0 0
1963 75 102 66679 10795 0.162 1.25 5.15 0.377 0 0 0
1968 83 113 90021 16757 0.186 1.32 5.47 0.403 0 0 0
Fibreboard packing cases 1954 27 43 38845 8289 0.213 0.94 4.07 0.367 0 0 0
1958 26 45 46420 8211 0.177 1.03 4.28 0.361 0 0 0
1963 46 77 75703 13950 0.184 0.92 3.79 0.377 0 0 0
1968 49 106 112448 18763 0.167 1.18 4.90 0.403 0 0 0
Composite containers of board 1963 10 12 5046 905 0.179 0.64 4.68 0.377 0 0 0
and metal 1968 8 9 6120 1250 0.204 0.81 3.04 0.403 0 0 0

Appendix B
Packaging products of paper and 1963 18 22 12122 3338 0.275 0.75 5.54 0.377 0 0 0
associated materials other than 1968 22 36 28625 4477 0.156 0.99 5.98 0.403 0 0 0
bags and sacks
Notepaper, envelopes, and 1954 38 50 23205 5164 0.223 0.50 2.17 0.367 0 0 1

Data Sets
boxed stationery, plain 1958 40 58 31337 6112 0.195 0.52 1.97 0.361 0 0 1
postcards, and letter cards 1963 29 44 30927 7668 0.248 0.91 3.18 0.377 0 0 1
Magazines and periodicals 1954 104 146 107087 23731 0.222 0.43 1.77 0.838 1 0 0
1958 91 128 127585 24462 0.192 0.66 2.22 0.838 1 0 0
1963 89 149 151136 38114 0.252 0.69 2.70 0.835 1 0 0
1968 101 170 149372 37386 0.250 0.75 3.67 0.793 1 0 0
Books 1954 127 139 54363 12424 0.229 0.35 2.10 0.838 1 0 0
1958 131 151 70757 15928 0.225 0.48 2.44 0.838 1 0 0
1963 113 149 89774 27755 0.309 0.76 3.91 0.835 1 0 0
Stereotyping, electrotyping, and 1954 132 167 15780 3569 0.226 0.13 1.95 0.838 0 0 1
engraving 1958 141 175 20284 4431 0.218 0.16 2.32 0.838 0 0 1
1963 141 198 25790 5856 0.227 0.20 2.89 0.835 0 0 1
Cellular rubber products 1954 7 9 11450 1369 0.120 2.14 4.21 0.304 0 0 0
1958 9 12 11155 1056 0.095 1.58 5.72 0.309 0 0 0
1963 11 16 15573 2260 0.145 1.18 3.96 0.344 0 0 0
Rubber products other than 1963 145 223 138082 28431 0.206 0.75 3.00 0.344 0 0 0
tyres, hose and tubing, and 1968 133 213 156152 38990 0.250 0.94 3.70 0.289 0 0 0
belting
Linoleum, plastic floor covering, 1954 22 29 50166 8403 0.167 2.12 3.97 0.304 0 0 1
leather cloth, etc. (MLH 492) 1958 21 31 47266 8095 0.171 2.25 4.75 0.309 0 0 1
1963 29 44 77154 17841 0.231 2.27 5.00 0.344 0 0 1
1968 21 31 71197 17027 0.239 3.81 6.97 0.289 0 0 1
1973 23 37 91327 25859 0.283 3.72 8.51 0.308 0 0 1

505
Table B6 (continued)

506
Industry Year NFIRMS NPLANTS SS PROFIT PCM K/N K/L UNION ADTYPE RDTYPE CHANGE
Brushes and brooms (MLH 493) 1954 116 143 20781 3529 0.170 0.13 1.51 0.304 1 0 0
1958 85 106 19241 4574 0.238 0.18 1.84 0.309 1 0 0
1963 66 85 20905 4226 0.202 0.25 2.23 0.344 1 0 0
1968 50 70 20913 4159 0.199 0.37 3.00 0.289 1 0 0
1973 60 70 24061 5528 0.230 0.45 3.66 0.308 1 0 0
Toys and games 1954 107 140 35143 6665 0.190 0.22 1.30 0.304 1 0 0
1958 87 117 35248 7988 0.227 0.29 1.69 0.309 1 0 0
1963 69 108 39519 10791 0.273 0.40 2.05 0.344 1 0 0
1968 64 104 57697 17575 0.305 0.66 2.55 0.289 1 0 0
Children’s and invalids’ 1954 23 25 5818 949 0.163 0.08 0.58 0.509 1 0 0
carriages 1958 17 19 4898 802 0.164 0.12 0.80 0.499 1 0 0
1963 21 27 7484 1390 0.186 0.15 1.06 0.528 1 0 0
1968 13 19 8401 1962 0.234 0.33 1.84 0.567 1 0 0
Pens and mechanical pencils 1954 28 35 7560 2110 0.279 0.36 2.93 0.304 2 0 0
1958 18 23 9454 3147 0.333 0.46 2.36 0.309 2 0 0
1963 14 18 9709 3428 0.353 0.56 2.86 0.344 2 0 0
Lead pencils, crayons, etc. 1958 12 12 2428 623 0.257 0.24 1.94 0.309 1 0 1
1963 11 12 2512 614 0.244 0.34 2.97 0.344 1 0 1
Office machinery requisites 1954 18 26 8343 2078 0.249 0.27 1.73 0.304 0 0 0

Appendix B
1958 19 25 10283 2118 0.206 0.29 1.37 0.309 0 0 0
1963 18 26 13234 2909 0.220 0.46 2.17 0.344 0 0 0
1968 21 28 19115 5384 0.282 0.68 3.84 0.289 0 0 0
Plastics products (MLH 496) 1954 218 258 45274 8217 0.182 0.22 1.95 0.304 0 0 0

Data Sets
1958 234 295 65068 11416 0.175 0.25 2.14 0.309 0 0 0
1963 377 491 145292 33741 0.232 0.32 2.63 0.344 0 0 0
1968 472 625 251331 62572 0.249 0.57 4.29 0.289 0 0 0
1973 645 722 394152 97511 0.247 0.85 6.26 0.308 0 0 0
Musical instruments 1954 51 63 5719 953 0.167 0.08 1.27 0.304 0 0 0
1958 39 50 5549 980 0.177 0.08 1.10 0.309 0 0 0
1963 30 42 6829 1684 0.247 0.08 0.83 0.344 0 0 0
1968 30 48 8285 1858 0.224 0.12 1.68 0.289 0 0 0
1973 34 52 13468 3323 0.247 0.14 1.85 0.308 0 0 0

Notes: The classification according to MLH industries follows the 1968 Standard Industrial Classification, unless otherwise stated.
NFIRMS is the number of firms employing at least 25 persons.
NPLANTS is the number of plants employing at least 25 persons.
SS is sales revenue (in £1000) deflated by the general producer price index.
PROFIT is the net value of output minus wages and salaries (in £1000) deflated by the general producer price index.
PCM is the price-cost margin, i.e., net value of output minus wages and salaries divided by sales revenue.
K is the value of capital stock in 1980 prices (in £ million) (my estimate; see text for details).
L is employment (in 1000).
N is the number of plants with at least 25 employees.
UNION is union density (i.e., the number of unionized employees over the total number of employees) of the corresponding industry group.
ADTYPE takes the value 0 for an industry with typical advertising-sales ratio lower than 1%, 1 for an industry with typical advertising-sales
ratio between 1% and 2%, and 2 for an industry with typical advertising-sales ratio higher than 2%.
RDTYPE takes the value 0 for an industry with typical R&D-sales ratio lower than 1%, 1 for an industry with typical R&D-sales ratio between
1% and 2%, and 2 for an industry with typical R&D-sales ratio higher than 2%.
CHANGE takes the value 1 for an industry with a change in competition regime and 0 otherwise.

507
This page intentionally left blank
Glossary of British Terms

Appendix A

UK US
Aluminium Aluminum
Animal feeding stuffs Animal foodstuffs
Children’s carriages Baby carriages
Electric cookers Electric stoves
Fibreboard packing cases Fiberboard packing cases
Gelatine Gelatin
Glass fibres Glass fibers
Gramophone records Phonograph records
Jewellery Jewelry
Lifts Elevators
Made-up textiles Fully manufactured textiles
Man-made fibres Man-made fibers
Pedal cycles Bicycles
Perry Fermented pear juice
Petrol Gasoline
Self-raising flour Self-rising flour
Slide fasteners Zippers
Spectacles Eyeglasses
Trams Streetcars
Tyres Tires
Wagons Railway freight cars
510 Glossary

Whisky Whiskey
Woollen Woolen

Appendix B

UK US
Accumulators Storage batteries
Aluminium Aluminium
Carbons Carbon paper
Children’s carriages Baby carriages
Electric cookers Electric stoves
Engineers’ and mechanicians’ goods Engineers’ and mechanics’ goods
Fibreboard packing cases Fiberboard packing cases
Gelatine Gelatin
Glass fibres Glass fibers
Gramophone records Phonograph records
Gramophones Phonographs
Ice lollies Ice pops
Invalids’ carriages Wheelchairs
Jewellery Jewelry
Lifts Elevators
Made-up textiles Fully manufactured textiles
Man-made fibres Man-made fibers
Perry Fermented pear juice
Potato crisps Potato chips
Pyjamas Pajamas
Sanitary towels Sanitary napkins
Spectacles Eyeglasses
Tyres Tires
Whisky Whiskey
Woollen Woolen
References

Publications used only as data sources are detailed in the text and are not included here.

Acs, Z. J., and D. B. Audretsch (1988). Innovation in large and small firms: An empir-
ical analysis. American Economic Review, 78: 678–690.
Aghion, P., M. Dewatripont, and P. Rey (1999). Competition, financial discipline and
growth. Review of Economic Studies, 66: 825–852.
Aghion, P., and P. Howitt (1997). A Schumpeterian perspective on growth and com-
petition. In D. M. Kreps and K. F. Wallis, eds., Advances in Economics and Econometrics:
Theory and Applications. Seventh World Congress. Vol. 2. Cambridge: Cambridge Uni-
versity Press.
Aghion, P., and P. Howitt (1998). Endogenous Growth Theory. Cambridge, MA: M.I.T.
Press.
Aghion, P., and M. Schankerman (2000). A Model of Market Enhancing Infrastructure.
CEPR Discussion Paper no. 2462. London: Centre for Economic Policy Research.
Armstrong, A., and A. Silberston (1965). Size of plant, size of enterprise and concen-
tration in British manufacturing industry 1935–1958. Journal of the Royal Statistical
Society, series A, 128: 395–420.

Asch, P., and J. J. Seneca (1976). Is collusion profitable? Review of Economics and Sta-
tistics, 58: 1–12.

Baden-Fuller, C. W. F. (1980). The Economics of Private Brands with Special Reference to


the Domestic Appliance Industries of the UK and USA. Ph.D. thesis, London School of
Economics.
Bain, G. S., and R. Price (1980). Profiles of Union Growth: A Comparative Statistical Por-
trait of Eight Countries. Oxford: Basil Blackwell.
Bertschek, I. (1995). Product and process innovation as a response to increasing
imports and foreign direct investment. Journal of Industrial Economics, 43: 341–357.
Bittlingmayer, G. (1985). Did antitrust policy cause the great merger wave? Journal of
Law and Economics, 28: 77–118.
Blundell, R., R. Griffith, and J. Van Reenen (1995). Dynamic count data models of
technological innovation. Economic Journal, 105: 333–344.
512 References

Board of Trade (1944). Survey of International Cartels. London: Board of Trade.


Board of Trade (1946). Survey of Internal Cartels. London: Board of Trade.
Board of Trade (1949). Report of the Committee on Resale Price Maintenance. London:
H.M.S.O.
Board of Trade (1949–1952). Monopolies and Restrictive Practices (Inquiry and Control)
Act, 1948: Annual Report by the Board of Trade. London: H.M.S.O.
Board of Trade (1953–1956). Monopolies and Restrictive Practices Acts, 1948 and 1953:
Annual Report by the Board of Trade. London: H.M.S.O.
Boone, J. (2000). Competitive pressure: The effects of investment in product and process
innovation. RAND Journal of Economics, 31: 549–569.

Bound, J., C. Cummins, Z. Griliches, B. H. Hall, and A. Jaffe (1984). Who does R&D
and who patents? In Z. Griliches, ed., R&D, Patents, and Productivity. Chicago: Uni-
versity of Chicago Press.
Bowley, A. L. (1924). The Mathematical Groundwork of Economics. Oxford: Oxford Uni-
versity Press.
Brander, J. A., and B. J. Spencer (1985). Tacit collusion, free entry and welfare. Journal
of Industrial Economics, 33: 277–294.
Bresnahan, T. F., and P. C. Reiss (1991). Entry and competition in concentrated markets.
Journal of Political Economy, 99: 977–1009.
Broadberry, S. N., and N. F. R. Crafts (1996). British economic policy and industrial
performance in the early post-war years. Business History, 38: 65–91.
Broadberry, S. N., and N. F. R. Crafts (2000). Competition and Innovation in 1950’s
Britain. Working Paper in Economic History no. 57. London School of Economics.
Brock, W. A., and J. A. Scheinkman (1985). Price setting supergames with capacity
constraints. Review of Economic Studies, 52: 371–382.

Burn, D., ed. (1958). The Structure of British Industry. 2 vols. Cambridge: Cambridge
University Press.

Cameron, A. C., and P. K. Trivedi (1998). Regression Analysis of Count Data. Cambridge:
Cambridge University Press.
Caves, R. E. (1988). Trade exposure and changing structures of US manufacturing
industries. In A. M. Spence and H. A. Hazard, eds., International Competitiveness.
Cambridge, MA: Ballinger.
Chevalier, J. A. (1995). Capital structure and product market competition: Empirical
evidence from the supermarket industry. American Economic Review, 85: 415–435.
Cohen, W. (1995). Empirical studies of innovative activity. In P. Stoneman, ed., Hand-
book of the Economics of Innovation and Technological Change. Oxford: Blackwell.
Conyon, M., and S. Machin (1991). Market structure and the empirical specification of
price-cost margins. Economics Letters, 35: 227–231.
Corley, T. A. B. (1966). Domestic Electrical Appliances. London: Jonathan Cape.
References 513

Cowling, K., and M. Waterson (1976). Price-cost margins and market structure. Eco-
nomica, 43: 267–274.

Crepon, B., and E. Duguet (1997). Research and development, competition and inno-
vation: Pseudo-maximum likelihood and simulated maximum likelihood methods
applied to count data models with heterogeneity. Journal of Econometrics, 79: 355–378.
Curry, B., and K. D. George (1983). Industrial concentration: A survey. Journal of
Industrial Economics, 31: 203–255.
Cyert, R. M., and M. H. deGroot (1973). An analysis of cooperation and learning in a
duopoly context. American Economic Review, 63: 24–37.
d’Aspermont, C., and M. Motta (2000). Tougher price competition or lower concen-
tration: A trade-off for antitrust authorities? In G. Norman and J.-F. Thisse, eds., Market
Structure and Competition Policy: Game-Theoretic Approaches. Cambridge: Cambridge
University Press.
Dasgupta, P., and J. Stiglitz (1980). Industrial structure and the nature of innovative
activity. Economic Journal, 90: 266–293.
Davidson, C., and R. Deneckere (1990). Excess capacity and collusion. International
Economic Review, 31: 521–541.
Davies, S. (1980). Minimum efficient size and seller concentration: An empirical prob-
lem. Journal of Industrial Economics, 28: 287–301.
deMelo, J., and S. Urata (1986). The influence of increased foreign competition on
industrial concentration and profitability. International Journal of Industrial Organi-
zation, 4: 287–304.
Demsetz, H. (1973). Industry structure, market rivalry, and public policy. Journal of
Law and Economics, 16: 1–9.

Dick, A. R. (1996a). Identifying contracts, combinations and conspiracies in restraint of


trade. Managerial and Decision Economics, 17: 203–216.

Dick, A. R. (1996b). When are cartels stable contracts? Journal of Law and Economics, 39:
241–283.
Director General of Fair Trading (DGFT) (1981). Annual Report of the Director General of
Fair Trading for the Period January 1980 to December 1980. London: H.M.S.O.
Dixit, A. (1979). A model of duopoly suggesting a theory of entry barriers. Bell Journal
of Economics, 10: 20–32.
Dixit, A., and J. Stiglitz (1977). Monopolistic competition and optimal product diver-
sity. American Economic Review, 67: 297–308.
Edgeworth, F. Y. (1881). Mathematical Psychics: An Essay on the Application of Mathe-
matics to the Moral Sciences. London: Kegan Paul.
Eissa, N. (1995). Taxation and Labour Supply of Married Women: The Tax Reform Act
of 1986 as a Natural Experiment. NBER Working Paper no. 5023. Cambridge, MA:
National Bureau of Economic Research.

Eissa, N., and J. B. Liebman (1996). Labour supply response to the earned income tax
credit. Quarterly Journal of Economics, 111: 605–637.
514 References

Elliott, D. C., and J. D. Gribbin (1977). The abolition of cartels and structural change in
the United Kingdom. In A. P. Jacquemin and H. W. deJong, eds., Welfare Aspects of
Industrial Markets. Leiden: Martinus Nijhoff.
Evely, R., and I. M. D. Little (1960). Concentration in British Industry: An Empirical Study
of the Structure of Industrial Production, 1935–51. Cambridge: Cambridge University
Press.
Farber, S. (1981). Buyer market structure and R&D effort: A simultaneous-equations
model. Review of Economics and Statistics, 63: 336–345.
Fershtman, C., and N. Gandal (1994). Disadvantageous semicollusion. International
Journal of Industrial Organisation, 12: 141–154.
Fershtman, C., and E. Muller (1986). Capital investments and price agreements in
semicollusive markets. RAND Journal of Economics, 17: 214–226.

Fershtman, C., and A. Pakes (2000). A dynamic oligopoly with collusion and price
wars. RAND Journal of Economics, 21: 207–236.

Fraas, A. G., and D. F. Greer (1977). Market structure and price collusion: An empirical
analysis. Journal of Industrial Economics, 26: 21–44.
Freyer, T. (1992). Regulating Big Business: Antitrust in Great Britain and America, 1880–
1990. Cambridge: Cambridge University Press.
Friedman, J. W., and J.-F. Thisse (1993). Partial collusion fosters minimum product
differentiation. RAND Journal of Economics, 24: 631–645.
Friedman, J. W., and J.-F. Thisse (1994). Sustainable collusion in oligopoly with free
entry. European Economic Review, 38: 271–283.
Fudenberg, D., and J. Tirole (1991). Game Theory. Cambridge, MA: M.I.T. Press.
Gasmi, F., J. J. Laffont, and Q. Vuong (1992). Econometric analysis of collusive behavior
in a soft-drink market. Journal of Economics and Management Strategy, 1: 277–312.
Geroski, P. A. (1990). Innovation, technological opportunity, and market structure.
Oxford Economic Papers, 42: 586–602.
Geroski, P. A. (1991). Market Dynamics and Entry. Oxford: Basil Blackwell.
Geroski, P. A. (1994). Market Structure, Corporate Performance and Innovative Activity.
Oxford: Clarendon Press.
Geroski, P. A., and R. Pomroy (1990). Innovation and the evolution of market struc-
ture. Journal of Industrial Economics, 38: 299–314.
Guenault, P. H., and J. M. Jackson (1974). The Control of Monopoly in the United Kingdom.
2nd ed. London: Longman.
Häckner, J. (1994). Collusive pricing in markets for vertically differentiated products.
International Journal of Industrial Organisation, 12: 155–177.
Hannah, L. (1979). Mergers, cartels and concentration: Legal factors in the US
and European experience. In N. Horn and J. Kocka, eds., Law and the Formation of the
Big Enterprises in the 19th and Early 20th Centuries. Göttingen: Vandenhoeck &
Ruprecht.
References 515

Hannah, L., and J. A. Kay (1977). Concentration in Modern Industry: Theory, Measurement
and the UK Experience. London: Macmillan.

Harrington, J. E. (1989). Collusion and predation under (almost) free entry. Interna-
tional Journal of Industrial Organization, 7: 381–401.
Harrington, J. E. (1991). The joint profit maximum as a free-entry equilibrium outcome.
European Economic Review, 35: 1087–1101.
Hart, P. E. (1985). Recent Trends in Concentration in British Industry. NIESR Discussion
Paper no. 82. London: National Institute of Economic and Social Research.
Hart, P. E., and R. Clarke (1980). Concentration in British Industry 1935–75. Cambridge:
Cambridge University Press.
Hart, P. E., and E. Morgan (1977). Market structure and economic performance in the
United Kingdom. Journal of Industrial Economics, 25: 177–193.
Hart, P. E., and S. J. Prais (1956). The analysis of business concentration: A statistical
approach. Journal of the Royal Statistical Society, series A, 119: 150–191.
Hart, P. E., M. A. Utton, and G. Walshe (1973). Mergers and Concentration in British
Industry. Cambridge: Cambridge University Press.
Hatch, J. H. (1970). Competition in the British White Goods Industry. Ph.D. thesis, Cam-
bridge University.
Hatch, J. H. (1972). Stop-go and the domestic appliance industry: A case study. Bulletin
of the Oxford University Institute of Economics and Statistics, 34: 345–358.
Hausman, J., B. H. Hall, and Z. Griliches (1984). Econometric models for count data
with an application to the patents-R&D relationship. Econometrica, 52: 909–938.

Hay, G. A., and D. Kelley (1974). An empirical survey of price fixing conspiracies.
Journal of Law and Economics, 17: 13–38.
Heath, J. B. (1961). Restrictive practices and after. Manchester School of Economic and
Social Studies, 29: 173–202.
Heath, J. (1963). Still Not Enough Competition? Business Restrictive Practices Re-examined.
2nd ed. London: Institute of Economic Affairs.
Hsiao, C. (1986). Analysis of Panel Data. Cambridge: Cambridge University Press.
Hunter, A. (1966). Competition and the Law. London: George Allen.

Ireland, N. J. (1987). Product Differentiation and Non-Price Competition. Oxford: Blackwell.


Jacquemin, A., E. deGhellinck, and C. Huveneers (1980). Concentration and profit-
ability in a small open economy. Journal of Industrial Economics, 29: 131–144.
Jacquemin, A., T. Nambu, and I. Dewez (1981). A dynamic analysis of export cartels:
The Japanese case. Economic Journal, 91: 685–696.

Jehiel, P. (1992). Product differentiation and price collusion. International Journal of


Industrial Organization, 10: 633–641.

Kamien, I., and N. L. Schwartz (1982). Market Structure and Innovation. Cambridge:
Cambridge University Press.
516 References

Kessides, I. N. (1990). Market concentration, contestability, and sunk costs. Review of


Economics and Statistics, 72: 614–622.

Kitchin, P. D. (1976). Effective rates of protection for United Kingdom manufacturing


in 1963 and 1968. In M. J. Artis and A. R. Nobay, eds., Essays in Economic Analysis: The
Proceedings of the Association of University Teachers of Economics, Sheffield 1975. Cam-
bridge: Cambridge University Press.
Klepper, S. (1996). Entry, exit, growth, and innovation over the product life cycle.
American Economic Review, 86: 562–583.
Klepper, S., and K. Simons (1997). Technological extinctions of industrial firms: An
enquiry into their nature and causes. Industrial and Corporate Change, 6: 379–460.
Kuenne, R. E. (1980). Duopoly reaction functions under crippled optimisation regimes.
Oxford Economic Papers, 32: 224–240.

Kuipers, J. D. (1950). Resale Price Maintenance in Great Britain: With Special Reference to
the Grocery Trade. Wageningen: N. V. Drukkerij ‘‘Vada.’’

Levin, R. C., and P. C. Reiss (1988). Cost-reducing and demand-creating R&D with
spillovers. RAND Journal of Economics, 19: 538–556.
Lunn, J. (1986). An empirical analysis of process and product patenting: A simultaneous-
equation framework. Journal of Industrial Economics, 34: 319–330.
Lydall, H. F. (1958). Aspects of competition in manufacturing industry. Bulletin of the
Oxford University Institute of Statistics, 20: 319–337.
Lyons, B., and C. Matraves (1996). Industrial concentration. In S. Davies, B. Lyons, et
al., Industrial Organization in the European Union: Structure, Strategy and the Competitive
Mechanism. Oxford: Clarendon Press.
MacKinnon, J. G., and H. White (1985). Some heteroskedasticity-consistent covariance
matrix estimators with improved finite sample properties. Journal of Econometrics, 29:
305–325.

Mansfield, E., J. Rapoport, J. Schnee, S. Wagner, and M. Hamburger (1971). Research


and Innovation in the Modern Corporation. London: Macmillan.
Martin, S. (1979a). Advertising, concentration, and profitability: The simultaneity
problem. Bell Journal of Economics, 10: 639–647.
Martin, S. (1979b). Entry barriers, concentration, and profits. Southern Economic Journal,
46: 471–488.
Martin, S. (2002). Advanced Industrial Economics. 2nd ed. Oxford: Blackwell.
Matraves, C. (1992). Endogenous Sunk Costs, Industry Size and Market Structure: A Four
Country Test. Discussion Paper no. 9224. Economics Research Centre, University of
East Anglia (Norwich).
Maunder, P. (1972). Price leadership: An appraisal of its character in some British
industries. Business Economist, 4: 132–140.
Mayers, J. B. (1963). Management and the British domestic electric appliance industry.
Journal of Industrial Economics, 12: 20–32.
References 517

Mercer, H. (1995). Constructing a Competitive Order: The Hidden History of British Anti-
trust Policies. Cambridge: Cambridge University Press.

Mercer, H. (1998). The Abolition of Resale Price Maintenance in Britain in 1964: A Turning
Point for British Manufacturers? Working Paper in Economic History no. 39. London
School of Economics.
Meyer, B. D. (1995). Natural and quasi-experiments in economics. Journal of Business
and Economic Statistics, 13: 151–161.
Milward, A. S., and G. Brennan (1996). Britain’s Place in the World: A Historical Inquiry
into Import Controls, 1945–1960. London: Routledge.
Monopolies and Restrictive Practices Commission (1955). Collective Discrimination: A
Report on Exclusive Dealing, Collective Boycotts, Aggregated Rebates and Other Discrim-
inatory Trade Practices. London: H.M.S.O.

Monopolies and Restrictive Practices Commission (1957). Report on the Supply and
Exports of Electrical and Allied Machinery and Plant. London: H.M.S.O.

Monopolies Commission (1963). Report on the Supply of Electrical Equipment for Mechan-
ically Propelled Land Vehicles. London: H.M.S.O.
Monopolies Commission (1973). Parallel Pricing: A Report on the General Effect on the
Public Interest of the Practice of Parallel Pricing. London: H.M.S.O.
Montagna, C. (1995). Monopolistic competition with firm-specific costs. Oxford Eco-
nomic Papers, 47: 318–328.
Morgan, A. D., and D. Martin (1975). Tariff reductions and UK imports of manufac-
tures: 1955–1971. National Institute Economic Review, no. 72: 38–54.
Motta, M. (1992). Co-operative R&D and vertical product differentiation. International
Journal of Industrial Organization, 10: 643–661.
Mukhopadhaya, A. K. (1985). Technological progress and change in market concen-
tration in the U.S., 1963–1977. Southern Economic Journal, 52: 141–149.
National Board for Prices and Incomes (NBPI) (1968). Prices of Secondary Batteries.
Report no. 61. London: H.M.S.O.
National Board for Prices and Incomes (NBPI) (1971). Prices, Profits and Costs in Food
Distribution. Report no. 165. London: H.M.S.O.
Nelson, R. L. (1959). Merger Movements in American Industry, 1895–1956. Princeton, NJ:
Princeton University Press.

Nicholson, R. J., and S. Gupta (1960). Output and productivity changes in British
manufacturing industry, 1948–1954: A study from Census of Production data. Journal
of the Royal Statistical Society, series A, 123: 427–459.
Nickell, S. (1995). The Performance of Companies: The Relationship Between the External
Environment, Management Strategies and Corporate Performance. Oxford: Blackwell.
Nocke, V. (2000). Monopolisation and Industry Structure. Nuffield College Economics
Paper 2000-w27. Oxford University.
Norman, G., and J.-F. Thisse (1996). Product variety and welfare under tough and soft
pricing regimes. Economic Journal, 106: 76–91.
518 References

O’Brien, D. P., W. S. Howe, D. M. Wright, and R. J. O’Brien (1979). Competition Policy,


Profitability and Growth. London: Macmillan.

O’Mahony, M., and N. Oulton (1990). Industry-Level Estimates of the Capital Stock in UK
Manufacturing, 1948–85. NIESR Discussion Paper no. 172. London: National Institute
of Economic and Social Research.
Overstreet, T. R. (1983). Resale Price Maintenance: Economic Theories and Empirical Evi-
dence. Washington, DC: Federal Trade Commission, Bureau of Economics.
Pakes, A., and M. Schankerman (1984). The rate of obsolescence of patents, research
gestation lags, and the private returns to research resources. In Z. Griliches, ed., R&D,
Patents, and Productivity. Chicago: University of Chicago Press.
Pavitt, K., ed. (1980). Technical Innovation and British Economic Performance. London:
Macmillan.

Phillips, A. (1972). An econometric study of price-fixing, market structure and per-


formance in British industry in the early 1950s. In K. Cowling, ed., Market Structure and
Corporate Behaviour: Theory and Empirical Analysis of the Firm. London: Gray-Mills.
Phlips, L. (1995). Competition Policy: A Game-Theoretic Perspective. Cambridge: Cam-
bridge University Press.
Pickering, J. F. (1966). Resale Price Maintenance in Practice. London: George Allen &
Unwin.
Pickering, J. F. (1974). The abolition of resale price maintenance in Great Britain. Oxford
Economic Papers, 26: 120–146.
Political and Economic Planning (PEP) (1957). Industrial Trade Associations: Activities
and Organisation. London: George Allen & Unwin.
Political and Economic Planning (1959). Tariffs and Trade in Western Europe. London:
George Allen & Unwin.
Pratten, C. F. (1971). Economies of Scale in Manufacturing Industry. Cambridge: Cam-
bridge University Press.
Registrar of Restrictive Trading Agreements (RRTA) (1961–1973). Report. London:
H.M.S.O.

Robinson, W. T., and J. Chiang (1996). Are Sutton’s predictions robust? Empirical
insights into advertising, R&D, and concentration. Journal of Industrial Economics, 44:
389–408.
Robson, M., J. Townsend, and K. Pavitt (1988). Sectoral patterns of production and use
of innovations in the UK: 1945–1983. Research Policy, 17: 1–14.
Rowley, C. K. (1966). The British Monopolies Commission. London: George Allen.
Salop, S. (1979). Monopolistic competition with outside goods. Bell Journal of Economics,
10: 141–156.
Sawyer, M. C. (1971). Concentration in British manufacturing industry. Oxford Eco-
nomic Papers, 23: 352–383.
Scherer, F. M. (1992). International High-Technology Competition. Cambridge, MA: Har-
vard University Press.
References 519

Scherer, F. M., A. Beckenstein, E. Kaufer, R. D. Murphy, and F. Bougeon-Maasen


(1975). The Economics of Multi-plant Operation: An International Comparisons Study.
Cambridge, MA: Harvard University Press.
Scherer, F. M., D. Harhoff, and J. Kukies (2000). Uncertainty and the size distribution
of rewards from innovation. Journal of Evolutionary Economics, 10: 175–200.
Scherer, F. M., and K. Huh (1992). R&D reactions to high-technology import competi-
tion. Review of Economics and Statistics, 74: 202–212.
Scherer, F. M., and D. Ross (1990). Industrial Market Structure and Economic Performance.
Boston: Houghton Mifflin.
Schmalensee, R. (1992). Sunk costs and market structure: A review article. Journal of
Industrial Economics, 40: 125–134.
Schott, K. (1976). Investment in private industrial research and development in Britain.
Journal of Industrial Economics, 25: 81–99.
Seade, J. (1980). On the effects of entry. Econometrica, 48: 479–489.
Secretary of State for Prices and Consumer Protection (SSPCP) (1979). A Review of
Restrictive Trade Practices Policy. London: H.M.S.O.
Selten, R. (1984). Are cartel laws bad for business? In H. Hauptmann, W. Krelle, and
K. C. Mosler, eds., Operations Research and Economic Theory: Essays in Honor of Martin
J. Beckmann. Berlin and Heidelberg: Springer-Verlag.

Shaked, A., and J. Sutton (1990). Multiproduct firms and market structure. RAND
Journal of Economics, 28: 605–628.
Shapley, L., and M. Shubik (1969). Price strategy oligopoly with product differen-
tiation. Kyklos, 22: 30–44.
Shaw, R. W., and C. J. Sutton (1976). Industry and Competition: Industrial Case Studies.
London and Basingstoke: Macmillan.
Shubik, M. (with R. Levitan) (1980). Market Structure and Behavior. Cambridge, MA:
Harvard University Press.
Simon, H. A., and C. P. Bonini (1958). The size distribution of business firms. American
Economic Review, 48: 607–617.
Slade, M. (1995). Product rivalry with multiple strategic weapons: An analysis of price
and advertising competition. Journal of Economics and Management Strategy, 4: 445–476.
Sleuwaegen, L., and H. Yamawaki (1988). The formation of the European common
market and changes in structure and performance. European Economic Review, 32:
1451–1475.

Spence, M. (1976). Product differentiation and welfare. American Economic Review,


Papers and Proceedings, 66: 407–414.
Suslow, V. (1991). Cartel contract duration: Empirical evidence from international
cartels. University of Michigan. Mimeo.
Sutton, J. (1991). Sunk Costs and Market Structure: Price Competition, Advertising, and the
Evolution of Concentration. Cambridge, MA: M.I.T. Press.
520 References

Sutton, J. (1996). Technology and market structure. European Economic Review, 40: 511–
530.

Sutton, J. (1997a). Game-theoretic models of market structure. In D. M. Kreps, and K.


F. Wallis, eds., Advances in Economics and Econometrics: Theory and Applications. Seventh
World Congress. Vol. 1. Cambridge: Cambridge University Press.
Sutton, J. (1997b). One smart agent. RAND Journal of Economics, 28: 605–628.
Sutton, J. (1998). Technology and Market Structure: Theory and History. Cambridge, MA:
M.I.T. Press.
Swann, D., D. P. O’Brien, W. P. J. Maunder, and W. S. Howe (1973). Competition in
British Industry: Case Studies of the Effects of Restrictive Practices Legislation. Department
of Economics, Loughborough University of Technology.
Swann, D., D. P. O’Brien, W. P. J. Maunder, and W. S. Howe (1974). Competition in
British Industry: Restrictive Practices Legislation in Theory and Practice. London: George
Allen & Unwin.

Symeonidis, G. (1996). Innovation, firm size and market structure: Schumpeterian


hypotheses and some new themes. OECD Economic Studies, 27: 35–70.
Symeonidis, G. (1998). The evolution of UK cartel policy and its impact on market
conduct and structure. In S. Martin, ed., Competition Policies in Europe. Amsterdam:
North-Holland.
Symeonidis, G. (1999a). Cartel stability in advertising-intensive and R&D-intensive
industries. Economics Letters, 62: 121–129.
Symeonidis, G. (1999b). In Which Industries is Collusion More Likely? Evidence from the
UK. CEPR Discussion Paper no. 2301. London: Centre for Economic Policy Research.
Symeonidis, G. (2000a). Price competition and market structure: The impact of cartel
policy on concentration in the UK. Journal of Industrial Economics, 48: 1–26.
Symeonidis, G. (2000b). Price competition, non-price competition, and market struc-
ture: Theory and evidence. Economica, 67: 437–456.
Symeonidis, G. (2000c). Price and nonprice competition with endogenous market
structure. Journal of Economics and Management Strategy, 9: 53–83.

Symeonidis, G. (2002). Cartel stability with multiproduct firms. International Journal of


Industrial Organization (forthcoming).
Tirole, J. (1988). The Theory of Industrial Organization. Cambridge, MA: M.I.T. Press.

Townsend, J., F. Henwood, G. Thomas, K. Pavitt, and S. Wyatt (1981). Science and
Technology Indicators for the UK: Innovations in Britain Since 1945. SPRU Occasional
Paper Series no. 16. University of Sussex.
Utterbach, J. M., and F. F. Suarez (1993). Innovation, competition, and industry struc-
ture. Research Policy, 22: 1–21.
Utton, M. A. (1970). Industrial Concentration. Harmondsworth, UK: Penguin.
Vickers, J. (1989). The nature of costs and the number of firms at Cournot equilibrium.
International Journal of Industrial Organization, 7: 503–509.
References 521

Vives, X. (1985). On the efficiency of Bertrand and Cournot equilibria with product
differentiation. Journal of Economic Theory, 36: 166–175.

Walshe, J. G. (1991). Industrial organisation and competition policy. In N. F. R. Crafts


and N. W. C. Woodward, eds., The British Economy Since 1945. Oxford: Clarendon
Press.
Waldman, D. M. (1982). A stationary point for the stochastic frontier likelihood. Journal
of Econometrics, 18: 275–279.
Ward, T. S. (1973). The Distribution of Consumer Goods: Structure and Performance. Cam-
bridge: Cambridge University Press.
Wilberforce, R. O., A. Campbell, and N. Elles (1966). The Law of Restrictive Trade Prac-
tices and Monopolies. 2nd ed. London: Sweet and Maxwell.
Wilson, T. (1962). Restrictive practices. In J. P. Miller, ed., Competition, Cartels and Their
Regulation. Amsterdam: North-Holland.
Wright, N. J. (1978). Product differentiation, concentration, and changes in concen-
tration. Review of Economics and Statistics, 60: 628–631.
Yamey, B. S. (1966a). Introduction: The main economic issues. In B. S. Yamey, ed.,
Resale Price Maintenance. London: Weidenfeld and Nicholson.

Yamey, B. S. (1966b). United Kingdom. In B. S. Yamey, ed., Resale Price Maintenance.


London: Weidenfeld and Nicholson.

Yarrow, G. K. (1985). Welfare losses in oligopoly and monopolistic competition. Jour-


nal of Industrial Economics, 33: 515–529.
Ziss, S. (1994). Strategic R&D with spillovers, collusion and welfare. Journal of Indus-
trial Economics, 42: 375–393.
This page intentionally left blank
Index

Abrasive Industries Association, 402 Aerial Ropeways Association, 364


Abrasives industry, 402 A.E.I. (Associated Electrical Industries),
Acs, Z. J., 247 219, 274n
Adhesives industry, 353–354 A.E.I.-Hotpoint, 213, 214
Advertising, 16, 181, 183–184, 276 Aerospace equipment manufacturing
and market structure as jointly endo- and repairing industry, 380
genous variables, 15–16, 149–150, Aggregated rebates, 23, 23n, 36, 38, 61n9,
156–161, 221–222, 328, 329–330 334. See also various industries
and price competition. See Price com- Aghion, P., 224, 275, 282n
petition, and advertising Agricultural Engineers Association, 360,
and product quality, 82, 95n7, 151 378
as sunk cost, 5–6, 91–92, 151–152 Agricultural Machine Parts Association,
data, 122–124, 175–177 361
in R&D-intensive industries, 237 Agricultural machinery (except tractors)
persuasive versus informative, 163 industry, 360–361
restrictions regarding, 35, 334 Agricultural Twine Manufacturers Asso-
theoretical models of, 151–172 ciation, 392
Advertising Association, 123, 124, 176, Aircraft Bolt and Nut Manufacturers
177n, 181 Association, 382
Advertising effectiveness, 6, 77, 83, 85, Aluminum and aluminum alloy indus-
92, 166, 166n9, 180, 200 try, 358–359
and TV advertising, 181–183 Animal foodstuffs industry, 343–
Advertising intensity, 76–77, 120, 122– 344
124, 173–174, 183–184 Apparel and Fashion Industry’s Asso-
and classification of industries, 122– ciation, 396, 397
124, 173–174 Architectural Cornish Granite Associa-
and collusive pricing, 79, 81, 82, 83, tion, 404
199–200 Area Electricity Boards, 372
econometric model for, 179–180, 202–205 Armstrong, A., 137
econometric results for, 205–211 Asbestos cement industry, 402
Advertising-intensive industries, 5–7, Asbestos industry, 395
15–16, 32, 35, 76, 149–150, 173–174, Asch, P., 74
178, 283–285, 328 Asphalt Roads Association, 404
constructing the samples of, 173–174, Associated Manufacturers of Domestic
175–176, 184–185, 198–199 Electric Appliances, 366, 376
Advertising-sales ratios, 76–77, 120, 122– Associated Manufacturers of Domestic
124, 173–177 Electric Cookers, 375
524 Index

Associated Manufacturers of Domestic Association of Scottish Galvanisers, 387


Electric Water Heaters, 376 Association of Shell Boilermakers, 367
Associated Transformer Manufacturers, Association of Steel Drum Manufac-
371 turers, 385
Association of British and Dominion Association of Tar Distillers, 349
Condensed Milk Manufacturers, 342 Anthracene Producers Committee, 349
Association of British Manufacturers of Naphthalene Producers Committee, 349
Milk Powder, 342 Pitch Pool, 349
Association of the British Pharmaceutical Association of Tinned Cream Manufac-
Industry, 350 turers, 342
Association of British Roofing Felt Manu- Association of Tube Rebuilders, 374
facturers, 403 Association of U.K. Plier Manufacturers,
Association of British Salted Fish Curers 381
and Exporters, 341 Association of Wood Block Manufac-
Association of Cleaning Waste Manu- turers, 407
facturers, 389 Association of Wood Wool Manufac-
Association of Corrugated Paper Makers, turers, 405
407n Associations. See Trade associations; see
Association of Crane Makers, 364 also names of various associations
Association of Dental Manufacturers and Asymmetric firms, 11–12, 19, 98–99,
Traders of the U.K., 369 162–163, 237–238, 281–283, 284, 326.
Association of Electric Steel Conduit See also Efficiency, differences between
Manufacturers, 357 firms; Quality, differences between
Association of Fish Meal Manufacturers, firms
343 and multiple equilibria, 11–12, 95–96
Association of Flax Canvas Weavers, 389 Audretsch, D. B., 247
Association of Folding Furniture Makers,
406 Bacon curing industry, 340
Association of Galvanised Steel Gutter Baden-Fuller, C. W. F., 217, 219
and Pipe Manufacturers, 387 Bain, G. S., 301
Association of Lace Paper Makers, 409 Bakery Allied Traders Association
Association of Lead Sellers, 360 (Imitation Cream Division), 342
Association of London Galvanisers, 387 Ball Clay Producers Federation, 337
Association of Makers of Newsprint, 407 Ball and roller bearing industry, 368
Association of Makers of Paper Ser- Band Saw Association, 381
viettes, 409 Barriers to entry. See Entry barriers
Association of Makers of Wood Free Basic Pig Iron Producers Association, 357
Papers, 407n Beckenstein, A., 191
Association of Manufacturers of Electric Bedding and soft furnishing industry,
Wiring Accessories, 378 406
Association of Manufacturers of Mohair Bedstead Fittings Association, 384
and Pile Floor Rugs and Mats, 393 Beer industry, 345
Association of Millers of Proprietary Beesley, M. E., 361
Brown Flour, 339 Benzole Producers, 349
Association of Musical Instrument Bertschek, I., 227
Industries, 414 Biscuit industry, 340
Association of Northern Master Electro- Bittlingmayer, G., 89
typers and Stereotypers, 411 Bituminous asphalt and emulsions,
Association of Paving Manufacturers, coated roadstone, and coated slag
405 industry, 404
Association of Plastic Cable Makers, 372 Black Bolt and Nut Manufacturers Asso-
Association of Scottish Bacon Curers, 340 ciation, 382
Index 525

Blanket Manufacturers Association, 391 British Bronze and Brass Ingot Manufac-
Blundell, R., 225, 248, 255n turers Association, 359
Board industry, 408 British Carton Association, 408
Board of Trade, 22, 26, 31, 49, 53, 54 British Celanese, 387
annual reports, 48–49, 50, 335. See also British Chemical Ware Manufacturers
various industries Association, 401
report on resale price maintenance. See British Cinema and Theatre Printers
Resale price maintenance, Lloyds’ Association, 411
Committee report on British Coking Industry Association, 347
surveys of cartels, 52, 54–55 British Concrete Pipe Association, 403
Boatbuilding industry, 378 British Constructional Steelwork Associ-
Boiler and boilerhouse plant industry, ation, 367
35, 367 British Cycle and Motor Cycle Industries
Bolt, nut, screw, rivet, etc. industry, 382– Association, 379
383 British Dextrine Manufacturers Associa-
Bonini, C. P., 98n11 tion, 344
Book Cloth Association, 389 British Elastic Braid Manufacturers
Book publishing industry, 32, 410 Association, 393
Boone, J., 224 British Electrical and Allied Manufac-
Bougeon-Maasen, F., 191 turers Association, 372
Bound, J., 232 British Electrical Industry Fair Trading
Bounds approach. See Market structure, Council, 378n
bounds approach to British Electro-ceramic Manufacturers
Bowley, A. L., 101n14 Association, 400
B.P. (British Petroleum), 347, 348 British Embroidery Transfer Manufac-
Bradford and District Waste Pullers turers Association, 411
Association, 391 British Fabric Federation, 390, 392
Branded Knitting Wool Association, British Federation of Master Printers,
390 411
Brander, J. A., 100n British Firework Manufacturers Safety
Bread industry, 339–340 Association, 354
Brennan, G., 38, 39n British Fluorspar Producers Association,
Bresnahan, T. F., 93n6 338
Briar Pipe Trade Association, 414 British Furniture Manufacturers Fed-
Brick industry, 399 erated Associations, 406
Bridge and Constructional Ironwork British Furniture Trade Confederation,
Association, 367 405, 406
Bridgewater Roofing Tile Manufacturers British Hacksaw Makers Association,
Association, 399 381, 382
Bright Bolt and Nut Manufacturers British Hard Metal Association, 381
Association, 382 British Hot Finished Tube Conference,
British Acetylene Association, 348 357
British Aluminium Foil Rollers Associa- British Impulse Clock Manufacturers
tion, 359 Association, 369
British Aluminium Hollow-ware Manu- British Iron and Steel Federation, 347,
facturers Association, 384 356
British Bacon Agents Association, 340 British Ironfounders Association, 358
British Bacon Curers Federation, 340 British Jacquard Engineers Association,
British Basic Slag, 353 363
British Bath Manufacturers Association, British Knitting Machine Builders Asso-
358 ciation, 363
British Bottle Association, 288 British Laboratory Ware Association, 401
526 Index

British Lock and Latch Manufacturers British wine, cider, and perry (fermented
Association, 385 pear juice) industry, 346
British Malleable Tube Fittings Associa- British Wire Netting Manufacturers
tion, 358 Association, 383n
British Mantle Association, 396 British Wrought Iron Association, 355
British Mat and Matting Manufacturers Broadberry, S. N., 37n12, 67–69, 227,
Association, 395 275
British Metal Spectacle Manufacturers Broadcast receiving and sound reproduc-
Association, 370 ing equipment industry, 374–375
British Metal Window Association, 385 Brock, W. A., 73
British Motor Trade Association, 379 Brushes and brooms industry, 412
British Non-Ferrous Metals Federation, Brush Wood Turners Association, 407
359 Buffalo Picker Manufacturers Associa-
British Nylon Spinners, 387, 388 tion, 396
British Oil Burner Manufacturers Asso- Buffer Rod Makers Association, 368
ciation, 368 Builders’ woodwork industry, 405
British Ophthalmic Lens Manufacturers Building materials and mineral products
Association, 370 industries, miscellaneous, 402–405
British Paint Advisory Council, 351 Built-up Roofing Council, 403
British Paper and Board Makers Associa- Burn, D., 36, 48, 338, 348, 350, 361, 382,
tion, 407, 408 399, 400
British Paper Bag Federation, 408 Bus Seat Frame Association, 379
British Paper Box Federation, 408 Butter industry, 341
British Paper Machine Felt Association, Butter Makers and Packers Association,
391 341
British Pump Manufacturers Association,
361 Cable and wire (insulated) industry, 36,
British Pyrotechnists Association, 354 372–373
British Radio Equipment Manufacturers Cadmium Copper and Bronze Associa-
Association, 375 tion, 359, 383
British Radio Valve Manufacturers Asso- Cameron, A. C., 255, 255n
ciation, 374 Campbell, A., 24n
British Refrigeration Association, 365, Can and metal box industry, 384
376 Can Manufacturers Association, 384
British Sanitary Earthenware Manufac- Canvas Hose Manufacturers Association,
turers Association, 400 389
British Sanitary Fireclay Association, Canvas sack and bag industry, 394
399 Capital intensity
British Starter Battery Association, 377 and collusion, 78, 79, 80, 83, 128
British Sulphate of Ammonia Federation, and concentration, 79, 128
353 measures of, 75–76
British Sulphate of Copper Association, Capital stock data, 75–76, 116–117, 120n,
348 300–301
British Surgical Trades Association, 369 Car and commercial vehicle industry,
British Synchronous Clock Conference, 379
369 Carpet industry, 393
British Tanning Extract Manufacturers Cartel policy, 1–2, 8, 12–13, 22–23, 24n,
Association, 352 33n8, 42, 146, 324, 328. See also Com-
British Tarpaviers Federation, 404 petition policy; Price competition;
British Teapot Manufacturers Associa- Restrictive Trade Practices Act (1956)
tion, 400 and industrial structure in the UK and
British Whiting Federation, 403 the US, 88–90, 278, 329
Index 527

and profitability in UK industry, 277– Coal Preparation Plant Association, 365


278, 329 Coal tar distillation industry, 349
Cartels, 17, 25, 36, 37, 38, 54–55, 106, Cocoa and chocolate industry, 342
107, 108, 295n4, 298, 324, 330. See also Coefficient of cooperation, 102n16
Collusion; Restrictive agreements; Coffee industry, 344
various industries Cohen, W., 225
data sources on, 46–50 Coke industry, 347
international, 38, 54–55, 334 Coleman, D. C., 387
Carton industry, 408 Collapsible Tube Association, 360
Cased Tube Association, 357 Collective discrimination, 23, 27, 61n9,
Case studies, 8, 211–221, 285–294 335. See also Aggregated rebates;
Cast Iron Axlebox Association, 358 Collective exclusive dealing; Resale
Cast Iron Chair Association, 358 price maintenance; various industries
Cast Iron Pressure Pipe Association, 358 MRPC report on, 23, 28, 48, 53, 334,
Cast Iron Segment Association, 358 335. See also various industries
Caves, R. E., 88 Collective exclusive dealing, 22, 23, 23n,
Cellulose film industry, 352 27, 28, 28n, 34, 36, 38, 61n9, 334. See
Cement industry, 65n13, 402 also various industries
Cement Makers Association, 402 Collusion, 13, 30, 33, 42, 42n, 65–66, 102,
Central Electricity Authority, 371, 372, 165. See also Cartels; Restrictive agree-
375 ments; various industries
Central Electricity Generating Board, and advertising intensity. See Advertis-
367, 371 ing intensity, and collusive pricing
Ceramic Printers Association, 411 and capital intensity. See Capital inten-
Chair Frames Association, 406 sity, and collusion
Cheddar and Caerphilly Cheese Makers and concentration. See Concentration,
Association, 341 and collusion
Cheese industry, 341–342 and entry, 17, 19, 36–37, 97, 281, 324,
Chemical (general, inorganic) industry, 330
348–349 and innovation. See Innovation, and
Chemical (general, organic) industry, 349 collusive pricing
Chemical (general, other than inorganic and prices, 39
and organic) industry, 349 and profitability, 37
Chemists Federation, 350 and R&D intensity. See R&D intensity,
Chemists Sundries Association, 369 and collusive pricing
Cheshire Cheese Makers Association, 341 and setup cost. See Setup cost, and
Chevalier, J. A., 135n44 collusion
Chiang, J., 88n, 193n22 determinants of, 13–14, 45–46, 71–83
Children’s (baby) carriages industry, 412 empirical studies of, 74
China and earthenware industry, 400 factors facilitating or hindering, 13–14,
China Clay Association, 337 45–46, 72–74
China Stone Association, 337 incidence across sectors, 69–71
Chloride, 377 tacit versus explicit, 50–51
Circular and Long Saw Association, 381, theoretical literature on, 72–74
382 variable, 74–75
Clarke, R., 106n19, 130, 220 Commercial Instruments Conference, 370
Clay Block Association, 399 Commission Recombers Association, 390
Clay industry, 337–338 Company of Scottish Cheesemakers, 341
Clock industry, 369 Competition, 19, 330. See also Price com-
Clothing Manufacturers Federation of petition; various industries
Great Britain, 396, 397 and efficiency, 10, 19, 37n12, 330
Clotted Cream Makers Association, 342 and innovation (literature on), 223–228
528 Index

Competition (cont.) Conyon, M., 311n18


effects of, 1, 8, 11–12, 14–18, 105, 173, Copper, brass, and other copper alloy
328 industry, 359
foreign, 38–39, 39n, 42, 130–131, 309– Corley, T. A. B., 212
310 Cork industry, 407
theoretical models of, 3–4, 6–7, 91–104, Corset industry, 397
151–172, 229–238. See also Game- Corsetry Manufacturers Association, 397
theoretic models Cotton Canvas Manufacturers Associa-
Competition data, 7, 13, 45, 46–56, 56– tion, 389
71, 113–114, 336 Cotton Velvet Council, 389
and classification of industries, 7, 13, Cotton Yarn Doublers Association, 388
45, 61–66, 67–69, 113–114, 336 Council of Iron Producers, 347
reliability of, 46–50, 50–56 Cournot-Nash equilibrium, 102, 165, 169,
Competition policy, 21, 33n8, 92. See also 170, 171
Cartel policy Courtaulds, 387, 388
implications for, 18–20, 331 Covered Conductors Association, 372
Compressor industry, 362 Cowling, K., 107n, 311
Computer industry, 375 Crafts, N. F. R., 37n12, 67–69, 227, 275
Concentration, 18, 19, 129–130, 331. See Crane industry, 364
also Market structure Cream industry, 342
adjustment to long-run equilibrium. See Crepon, B., 232
Market structure, adjustment to long- Cresylic Acid Refiners Committee, 349
run equilibrium Crown Cork, 384
and capital intensity. See Capital inten- Cummins, C., 232
sity, and concentration Curry, B., 106, 146
and collusion, 78, 79, 80, 81, 127–128, Cycle Trade Union, 380
185–187, 261–263 Cyert, R. M., 102n16
and market size. See Market size, and Cylinder and Refined Iron Association, 357
concentration Cylinder Lock Manufacturers Associa-
and price competition. See Price com- tion, 385
petition, and concentration
and setup cost. See Setup cost, and Dasgupta, P., 230n5
concentration d’Aspermont, C., 95n8
as a proxy for market power, 225 Dairy Engineers Association, 366
data, 105–110 Davidson, C., 224
econometric models for, 124–125, 129– Davies, S., 115
130, 132–138, 179–180, 187–188, 250– deGhellinck, E., 278
251, 264–265 deGroot, M. H., 102n16
econometric results for, 138–146, 188– deMelo, J., 278
198, 208–211, 265–274 Demsetz, H., 93n5
measurement of, 75, 106–109, 132. See Deneckere, R., 224
also Market structure, measures of Detergent industry, 351–352
Concrete Mixers Manufacturers Associa- Dewatripont, M., 224
tion, 363 Dewez, I., 74n17
Condensed milk industry, 342 Diary Publishers Association, 410
Conditions of sale, 34, 57, 59–60, 60n, 333 Dick, A. R., 74, 74n17
Conduit Fittings Manufacturers Associa- Difference-in-differences method, 8–10
tion, 357 limitations of, 9–10
Construction and earth-moving equip- validity of, 9–10, 84–85, 135–137, 203–
ment industry, 363 205, 208, 256–257, 306–307, 322. See
Conveyor, aerial ropeway, etc. industry, also Intensity of price competition,
364 exogeneity of
Index 529

Director General of Fair Trading, reports, prices, 287, 288


60n, 336, 340, 345, 402, 403, 404 profitability, 286, 287
Dixit, A., 95n7, 100 restrictive agreements, 286, 287, 371
Dog and cat food industry, 344 Restrictive Practices Court case, 286, 287,
Dress industries, miscellaneous, 398 371
Dress, lingerie, infants’ wear, etc. indus- Electrical ware industry, 400
try, 397 Electrolux, 217
Drop forging, stamping and pressing Electro-medical Trade Association, 375
industry, 384 Electronic valve, cathode ray tube, etc.
Dry Dock Owners and Repairers Central industry, 374
Council, 378 Electrotyping and Stereotyping Em-
Duguet, E., 232 ployers Federation, 411
Duomatic, 214 Elles, N., 24n
Dustbin Association, 385 Elliott, D. C., 89, 106n19, 329
Dyers and Finishers Association, 394n Employment data, 76, 119–120
Dyestuff industry, 352 Enamel Association, 399
Engineers’ small tool and gauge indus-
Earthenware Association, 400 try, 381
Edge Tool Manufacturers Association, English Butter Conference, 341
381 English China Manufacturers Associa-
Edgeworth, F. Y., 102n16 tion, 400
Efficiency English Electric, 219, 274n
and competition. See Competition, and English Joinery Manufacturers Associa-
efficiency tion, 405
differences between firms, 37, 93n5, 98– Entry, 3, 17, 19–20, 324, 325–326, 330
99, 281–283, 325. See also Asymmetric and collusion. See collusion, and entry
firms and zero-profit condition, 3, 14, 15, 19–
Eissa, N., 135n44 20, 87, 150, 280, 324, 325–326
Electric lamp industry, 36, 377 barriers, 3, 19, 88, 93n5, 96, 330, 334
Electric Lamp Manufacturers Associa- restrictions regarding, 36–37
tion, 377 Envelope Makers and Manufacturing
Electric Light Fittings Association, 377 Stationers Association, 409
Electric light fittings and wiring acces- Evely, R., 137
sories industry, 377–378 Exit, 14, 20, 87, 98, 147, 280, 313, 325, 326
Electric Sign Manufacturers Association, Excavator Makers Association, 363
378 Exogenous sunk cost industries, 5–6, 14,
Electrical appliance (domestic) industry, 17, 87, 91–92, 120, 124, 280, 285, 302,
211, 212, 375–376. See also Refrigerator 327–328
industry; Washing machine industry constructing the samples of, 120, 125–
Electrical equipment for motor vehicles, 127, 303–304
cycles, and aircraft industry, 376 Explosives industry, 354
Electrical machinery industry, 35, 61n8,
82, 370–372 Farber, S., 226
Electrical motor industry, 372 Federal Trade Commission, 121
Electrical power transformer industry, Federation of Bone Users and Allied
286–288, 371 Trades, 344, 353
demand growth, 287 Federation of British Carpet Manufac-
exit, 287 turers, 393
market structure, 287–288 Federation of Building Blocks Manufac-
mergers, 287 turers, 403
MRPC inquiry, 286, 287, 371 Federation of Engineers’ Sensitised
price competition, 286–287, 371 Material Manufacturers, 355
530 Index

Federation of Gelatine and Glue Manu- Footwear industry, 398


facturers (Hide, Gelatine and Glue Ford Motor, 361
Section), 353 Foundry Pig Iron Producers Association,
Federation of Light Metal Smelters, 358 357
Federation of Master Organ Builders, 414 Fraas, A. G., 74
Federation of Master Process Engravers, Freyer, T., 88–89
411 Friedman, J. W., 150, 281
Federation of Paper Tube Manufacturers, Frigidaire, 217
409 Fruit and vegetable products industry,
Federation of Reinforcement Fabric 343
Makers, 383n Fudenberg, D., 92
Federation of Wholesale and Multiple Fur industry, 396
Bakers, 339 Furniture and upholstery industry, 405
Fellmongery industry, 395–396
Fershtman, C., 73n, 150, 224 Galvanised Hollow-ware Association,
Fertiliser Manufacturers Association, 353 384
Fertilizer industry, 30, 353 Galvanised Tank Association, 385
Fiberboard packing case industry, 408 Game-theoretic models, 2, 3–4, 6–7, 327
Fibreboard Packing Case Manufacturers for advertising-intensive industries,
Association, 408 151–172
File Trade Association, 381 for exogenous sunk cost industries, 91–
Fine Art Trade Guild, 411 104
Fireclay Grate Back Association, 398 for R&D-intensive industries, 229–238
Fireclay goods industry, 399 Gandal, N., 150, 224
Firework industry, 354 Garage Equipment Association, 367
Firm numbers Gas appliance (domestic) industry, 386
and market size. See Market size, and Gas Area Boards, 386
firm/plant numbers Gas Council, 386
and price competition. See Price compe- Gasholder Makers Association, 367
tition, and firm/plant numbers Gas Meter Makers Conference, 370
constructing the samples for the analy- Gasmi, F., 149
sis of, 294–297, 303–304 G.E.C. (General Electric Co.), 219, 274n
data, 297 Gedges Drawback Hook Manufacturers
econometric model for, 302–303, 309– Association, 368
312 George, K. D., 106, 146
econometric results for, 312–313, 314– Generator and turbine industry, 370–371
315, 318, 319, 322–323 Geroski, P. A., 93n6, 225, 226n3, 227,
Fishing Boats Builders Association, 378 231n7, 241, 242, 242n15, 255, 275, 323
Fixed-effects model, 137–138, 188, 192, Glass Benders Association, 401
205, 255, 264–265, 312 Glass Bottle Association, 401
Flexible Back Bandsaw Manufacturers Glass container industry, 288–289, 401
Association, 381 demand growth, 288
Floor Quarry Association, 399 efficiency, 289
Flour confectionery industry, 340 market structure, 289
Flour industry, 338–339 mergers, 289
Fluid power equipment industry, 362 price competition, 288–289, 401
Flushing Cistern Association, 413 prices, 288–289
Food and drink processing machinery profitability, 289
industry, 366 restrictive agreements, 288, 401
Food industries, miscellaneous, 344 Restrictive Practices Court case, 288,
Food Machinery Association (Brewery 401
Equipment Section), 366 Glass industry, 400–402
Index 531

Glazed Cement Manufacturers Associa- High Speed Steel Alloys, 360


tion, 405 High Speed Steel Drill Rod Association,
Glazed and Floor Tile Manufacturers 381
Association, 400 High Speed Steel Tool Bit Association, 381
Glove industry, 398 Hook and Band Manufacturers Associa-
Grain milling industry, 338–339 tion, 387
Gramophone (phonograph) record and Hoover, 213, 214
tape recording industry, 374–375 Hosiery industry, 392
Granite Kerb Conference, 404 Howe, W. S., 21, 22n, 24n, 27, 28, 31, 36,
Greer, D. F., 74 37, 38, 39, 41–43, 43n, 51, 55, 81n24,
Greeting Card and Calendar Association, 89–90, 146, 228, 277–278, 285–292, 329,
411 335, 337, 339, 355, 357, 358, 359, 371,
Gribbin, J. D., 89, 329 372, 373, 377, 383, 385, 391, 393, 399,
Griffith, R., 225, 248, 255n 400, 401, 402, 412
Griliches, Z., 232, 255 Howitt, P., 224, 275
Guenault, P. H., 22n, 36 Hsiao, C., 137
Guild of Metal Perforators, 387 Huh, K., 226
Gupta, S., 111n28 Hunter, A., 24n, 25, 27, 28, 31
Huveneers, C., 278
Häckner, J., 73 Hydraulic Association, 362, 366
Hairdressing Manufacturers and Whole-
salers Association, 351 I.C.I. (Imperial Chemical Industries), 387,
Hall, B. H., 232, 255 388
Hamburger, M., 247 Ice cream industry, 342
Hand tool and implement industry, 381– Illuminating glassware industry, 401
382 Independent Cable Makers Association,
Hannah, L., 88–89, 106 372
Hard Fibre Cord and Twine Manufac- Industrial engine industry, 362–363
turers Association, 392 Industrial plant (including process plant)
Hard Fibre Rope Manufacturers Associa- industry, 367
tion, 392 Information agreements, 24, 31, 41, 42,
Hardwood and softwood industry, 405 52, 59, 334. See also various industries
Harhoff, D., 238n Innovation
Harrington, J. E., 281 and collusive pricing, 83, 252
Harris Tweed Industry Consultative and competition (literature on) 223–228
Committee, 391 and market structure as jointly endo-
Hart, P. E., 98n11, 106n19, 130, 215, genous variables, 16, 225–226, 228–
219n35, 220, 311, 311n18, 342, 346, 373, 229, 235–236, 274–275, 328
374, 390, 391 and price competition. See Price compe-
Hat, cap, and millinery industry, 397 tition, and innovation
Hatch, J. H., 212, 215, 216, 219 data, 240–250
Hausman, J., 232, 255 econometric model for, 250–251, 253–
Hay, G. A., 74 257
Heated Bolt Association, 382 econometric results for, 257–260
Heath, J. B., 38, 40–41 lags, 246–249
Heat-treated milk industry, 341 measures of, 238, 239, 249, 250, 253–254
Heavy Coil Spring Association, 385 theoretical model of, 229–238
Hebridean Spinners Advisory Com- Intensity of price competition, 1–2, 4–5,
mittee, 391 43, 92–93, 102–103, 153, 165–166, 230.
Henwood, F., 241, 244, 249 See also Price competition
High Conductivity Copper Association, and zero-profit condition, 94, 155, 233.
359, 383 See also Collusion, and entry
532 Index

Intensity of price competition (cont.) Lancashire Mechanical Cloth Manufac-


exogeneity of, 5, 9–10, 84–85, 99–100, turers Association, 391
103, 135–137, 163–164, 166, 203–205, Large Tube Association, 357
208, 237, 256–257, 306–307, 322 Lawn mower industry, 366
measurement of, 1–2, 229, 328 Law Stationers Association, 411
variable, 113–114, 133, 188, 202–203, Lead industry, 360
254, 264, 299, 310 Lead Sheet and Pipe Association, 360
Internal combustion engine industry, 362 Leather goods industry, 396
Ireland, N. J., 150 Leather (tanning and dressing) industry,
Iron and Steel Board, 338, 356, 357 395–396
Iron and steel (general) industry, 355– Legion Information Services, Ltd., 122–
357 124, 175–177, 213n
Iron castings industry, 358 Lentil Millers Association, 339
Iron ore industry, 338 Levin, R. C., 226
Liebman, J. B., 135n44
Jackson, J. M., 22n, 36 Lift (elevator) and escalator industry, 364
Jacquemin, A., 74n17, 278 Light Clothing and Allied Trades Associ-
Jaffe, A., 232 ation, 397
Jehiel, P., 150 Light Edge Tool and Allied Trades Asso-
Jewelry and precious metals industry, ciation, 381
384 Light Metal Founders Association, 359
Jute industry, 391 Lime and whiting industry, 403
Jute Sack and Bag Manufacturers Associ- Linear demand model, 100–104
ation, 394 Linear demand model with quality, 164–
171, 229n
Kamien, I., 232 Linen Sewing Thread Manufacturers
Kapok Processors Association, 395 Association, 388
Kaufer, E., 191 Linoleum Manufacturers Association,
Kay, J. A., 106 412
Kelley, D., 74 Linoleum, plastic floor covering, leather
Kessides, I. N., 115 cloth, etc. industry, 35n10, 412
Kiln Owners Association, 405 Little, I. M. D., 137
Kitchin, P. D., 39n, 131 Locked Coil Ropemakers Association,
Klepper, S., 226 383
Knife, tableware, etc. industry, 382 Locomotive and Allied Manufacturers
Knitted, netted, or crocheted clothing Association, 381
industry, 392 Locomotive and railway track equip-
Knitted, netted, or crocheted fabrics ment industry, 380–381
industry, 392 London Trade Typesetters Association,
Kuenne, R. E., 102n16 411
Kuipers, J. D., 33n9 London Wholesale Millinery Manufac-
Kukies, J., 238n turers Association, 397
Lubricating oils and greases industry,
Lace industry, 393 348
Lace Machine Builders and Allied Trades Lucas, 376, 377
Association, 363 Lunn, J., 226
Laffont, J. J., 149 Lydall, H. F., 52, 54, 54n
Laminated Railway Spring Association, Lyons, 345
386 Lyons, B., 96, 193n22
Lampshade Manufacturers Association,
413 Machine Knife Manufacturers Associa-
Lancashire Cheese Association, 341 tion, 382
Index 533

Machinery Belt Fastener Manufacturers and profitability/profits as jointly


Association, 386 endogenous variables, 17–18, 277–279,
Machinery (nonelectrical) industries, 280–285, 311, 319, 323–325, 327–328,
miscellaneous, 366–367 329
Machine Tool Trades Association, 361 bounds approach to, 11–12, 98
Machin, S., 311n18 endogeneity of, 3, 6–7, 93n5, 150, 225–
MacKinnon, J. G., 138, 188, 312 226
Made-up (fully manufactured) house- measures of, 178, 298. See also Concen-
hold textiles industry, 394 tration, measurement of
Made-up (fully manufactured) textiles theoretical models of, 3–4, 6–7, 91–104,
(of sailcloth, canvas, etc.) industry, 151–172, 229–238. See also Game-
394 theoretic models
Madras Manufacturers Association, 393 Martin, D., 39n, 131
Magnesite and Chrome Brickmakers Martin, S., 73, 88, 101n14
Association, 398 Mastic Asphalt Employers Federation,
Mains Cable Manufacturers Association, 404
372 Match industry, 354
Mains Cable Manufacturers Association Matraves, C., 96, 193n22
(Super Tension), 372 Maunder, P. J., 21, 22n, 24n, 27, 28, 31,
Malt industry, 345 36, 37, 38, 39, 41–43, 43n, 51, 55,
Maltsters Association of Great Britain, 81n24, 146, 228, 285–292, 329, 330, 335,
345 337, 339, 355, 357, 358, 359, 371, 372,
Man-made fiber industry, 387–388 373, 377, 383, 385, 391, 393, 399, 400,
Mansfield, E., 247 401, 402, 412
Manufactured fuel industry, 347 Mayers, J. B., 213, 216
Manufacturing industries, miscellaneous, Meat and fish products industry, 340–
414 341
Margarine industry, 344 Mechanical engineering industries,
Marine engineering industry, 378 miscellaneous, 368–369
Marine Traders Association, 378 Mechanical Handling Engineers Associa-
Market sharing, 22, 23, 27, 28, 34, 57, 60, tion, 364
61, 333, 334, 335. See also various indus- Mechanical handling equipment indus-
tries try, 364
Market size Media Expenditure Analysis, Ltd.
and concentration, 139, 192, 193, 265 (MEAL), 122–124, 175–177, 213n
and firm/plant numbers, 313 Medical and Surgical Plaster Makers
data, 111–113 Conference, 355
measures of, 110–111, 298–299 Medical, surgical, dental, and veterinary
Market structure, 18, 98n11. See also instrument and appliance industry,
Concentration 369–370
adjustment to long-run equilibrium, 97, Men’s and boys’ tailored outerwear
133–134, 280 industry, 396
and advertising as jointly endogenous Mercer, H., 22n, 25, 59n6
variables. See Advertising, and market Mergers, 14, 18, 87, 89–90, 97n, 98, 130,
structure as jointly endogenous 146, 278, 280, 309, 313, 325, 331
variables Merseyside Non-Ferrous Founders Asso-
and innovation as jointly endogenous ciation, 359
variables. See Innovation, and market Metal Bedstead Association, 384
structure as jointly endogenous Metal Box, 384
variables Metal-cutting and metal-forming
and price competition. See Price compe- machine tool industry, 361
tition, and market structure Metal finishing industry, 387
534 Index

Metal furniture industry, 384 Motta, M., 95n8, 230n5


Metal goods industries, miscellaneous, Mukhopadhaya, A. K., 226n3
385–387 Muller, E., 224
Metal hollow-ware industry, 384–385 Murex, 360
Metallic closures industry, 386 Murphy, R. D., 191
Metal Skewer Association, 386 Musical instrument industry, 414
Metal window and door frame industry, Music Publishers Association, 410
385
Meter Manufacturers Association, 370 Nambu, T., 74n17
Meyer, B. D., 9 Narrow fabrics industry, 393–394
Mica Trade Association, 405 Nash equilibrium, 155, 233
Mild Steel Wire Manufacturers Associa- National Association of Biscuit Manu-
tion, 383n facturers, 340
Milk and milk products industry, 341– National Association of British and Irish
342 Millers, 339
Milking Machine Manufacturers Associa- National Association of Creamery Pro-
tion, 361 prietors and Wholesale Dairymen, 342
Milk Marketing Board, 341 National Association of Dog Biscuit
Milk powder industry, 342 Manufacturers, 344
Millers Mutual Association, 339 National Association of Drop Forgers
Millinery Distributors Association, 397 and Stampers, 384
Millinery Guild, 397 National Association of Engravers and
Milling Cutter and Reamer Association, Die-Stampers, 411
381 National Association of Hematite Pig
Milward, A. S., 38, 39n Iron Makers, 357
Mineral oil refining industry, 347–348 National Association of Lift Makers, 364
Mineral Water Engineers Association, National Association of Manufacturers
366 and Distributors of Office Equipment,
Mining machinery industry, 365 406
Mining Rope Association, 383 National Association of Putty Manufac-
Monopolies and Restrictive Practices Act turers, 351
(1948), 22, 335 National Association of Shopfitters, 406
Monopolies and Restrictive Practices National Association of Upholstery Fibre
Commission (MRPC), 22–23, 24, 25, Processors, 406
27, 28, 51, 53, 335 National Bedding Federation, 406
reports, 22–23, 27, 28, 36, 37, 38, 47–48, National Board for Prices and Incomes,
50, 53, 60, 61n8, 61n9, 122, 334, 335. reports, 42, 122, 183, 408
See also various industries National Brassfoundry Association, 385,
Monopolies [and Mergers] Commission 387
(M[M]C), 25, 48, 335 National Caravan Council, 379
reports, 42, 47–48, 50, 122. See also National Clayware Federation, 399
various industries National Coal Board, 365
Montagna, C., 282 National Federation of Associated Paint,
Moquette Manufacturers Association, Colour and Varnish Manufacturers of
391 the U.K., 351
Morgan, A. D., 39n, 131 National Federation of Engineering and
Morgan, E., 311, 311n18 General Ironfounders, 358
Motorcycle, tricycle, and bicycle indus- National Federation of Retail News-
try, 379–380 agents, Booksellers and Stationers, 409
Motor vehicle industry, 379 National Federation of Scrap Iron, Steel
Motor vehicle parts and accessories and Metal Merchants, 356
industry, 379 National Firebrick Conference, 398
Index 535

National Forgemasters Association, 368, Office machinery industry, 364–365


384 Office of Fair Trading, 46, 106n19
National Health Service (NHS), 350 Ophthalmic Prescription Manufacturers
National Horticultural Pottery Manufac- Association, 370
turers Association, 399 Optical instrument industry, 370
National Hosiery Manufacturers Federa- Ordnance and small arms industry, 368
tion, 392 Oulton, N., 75, 76, 116–117, 120n, 300–
National Ingot Mould Association, 358 301
National Salt Glazed Pipe Manufacturers Overalls and men’s shirts, underwear,
Association, 399 etc. industry, 397
National Silica Brickmakers Association, Overstreet, T. R., 58
398
National Sulphuric Acid Association, 348 Packaging products (of paper) industry,
National Tile Fireplace Makers Associa- 408
tion, 404 Paint industry, 351
Natural experiments, 1, 8–10, 135n44, Pakes, A., 73n, 150, 224, 247
328 Paper and board manufactures indus-
Needle, pin, and other metal small-ware tries, miscellaneous, 409
industry, 386 Paper box industry, 408
Nelson, R. L., 89 Paper industry, 407–408
Newspaper Proprietors Association, 409 Parallel pricing, 42, 42n, 49
Newspaper publishing industry, 409 MMC report on, 42
Nicholson, R. J., 111n28 Patented Steel Wire Association, 383n
Nickel industry, 360 Patent Glazing Conference, 385
Nickell, S., 225 Patent pooling, 35–36, 82
Nocke, V., 98n10 Pavitt, K., 122, 241, 242, 244, 249
Nonferrous metal industries, miscellane- Pen and mechanical pencil industry, 413
ous, 360 Pencil industry, 413
Nonmetalliferous mining and quarrying Pencil Makers Conference, 413
industry, 338 Periodical Proprietors Association, 410
Norman, G., 95n8 Periodical publishing industry, 410
North East Coast Ship Repairers Associ- Permanent Magnet Association, 356
ation, 378 Permanent magnet industry, 35n11, 356–
North East Concrete Producers Associa- 357
tion, 403 Pesticide and disinfectant industry, 354
Northern Ireland Plate Glass Associa- Pharmaceutical chemicals industry, 350
tion, 401 Pharmaceutical preparations industry,
Northern Tobacco Manufacturers Associ- 32, 350–351
ation, 346 Phenol Producers Association, 349
North Wales Slate Quarries Association, Phillips, A., 74
337 Phlips, L., 280
Notification of inquiries, 34–35, 57, 60–61 Phosphate Rock Agency, 353
Photographic chemical materials indus-
O’Brien, D. P., 21, 22n, 24n, 27, 28, 31, try, 355
36, 37, 38, 39, 41–43, 43n, 51, 55, Photographic and document-copying
81n24, 89–90, 146, 228, 277–278, 285– equipment industry, 369
292, 329, 330, 335, 337, 339, 355, 357, Photo-litho Reproducers Association, 411
358, 359, 371, 372, 373, 377, 383, 385, Pickering, J. F., 59, 183
391, 393, 399, 400, 401, 402, 412 Pig iron industry, 357
O’Brien, R. J., 89–90, 277–278, 329 Pigment industry, 352–353
O’Mahony, M., 75, 76, 116–117, 120n, Pigskin Tanners Federation, 395
300–301 Pilkington Brothers, 400, 401
536 Index

Pin and Allied Trades Association, 386 and advertising, 15–16, 19, 149–150,
Plaited Cordage Association, 392 151–172, 178–179, 199–202, 205–211,
Plant numbers 221–222, 276, 328, 329–330
and market size. See Market size, and and concentration, 14–16, 18, 87–90,
firm/plant numbers 91–104, 127–129, 138–147, 149–150,
and price competition. See Price compe- 151–172, 178–179, 185–187, 188–198,
tition, and firm/plant numbers 208–211, 221–222, 228–229, 229–238,
constructing the samples for the analy- 261–264, 265–274, 274–275, 283, 327–
sis of, 294–297, 303–304 328, 329, 330, 331
data, 76, 117–119, 120n, 297 and firm/plant numbers, 17, 277–279,
econometric model for, 302–303, 309– 280, 283, 284–285, 304–309, 312–313,
312 314–315, 318, 319, 323–325, 327–328,
econometric results for, 312–313, 314– 330
315, 318, 319, 322–323 and innovation, 16, 19, 228–229, 229–
Plaster Ventilators Manufacturers Asso- 238, 252–253, 257–260, 274–276, 328,
ciation, 405 330, 331
Plastic products industry, 413 and market structure, 3. See also Price
Plastics Hardware Association, 413 competition, and concentration; Price
Plastic Spectacle Manufacturers Associa- competition, and firm/plant numbers
tion, 370 and profitability/profits, 3, 14, 15, 17–
Plate Glass Association, 401 18, 87, 146, 147, 277–279, 280–285,
Plate glass industry, 400–401 304–309, 313, 316–322, 323–325, 327–
Plywood and textured board industry, 328, 329
405 Price-fixing, 22, 23, 27, 28, 34, 35, 53, 57,
Pneumatic Metal Working Tool Associa- 58, 60, 61, 333, 334, 335. See also various
tion, 366 industries
Pneumatic Tool Association, 366 Price, R., 301
Polishes industry, 353 Prices and incomes policy, 42, 184, 309
Political and Economic Planning, 38, Primary battery industry, 376
39n Printing, bookbinding, and paper goods
survey of trade associations, 33n9, 38, making machinery industry, 365
48, 49–50, 60, 81, 335. See also various Printing (general) industry, 411
industries Printing ink industry, 354
Pomroy, R., 226n3 Procter and Gamble, 351
Portable Air Compressor Association, Product differentiation
362 horizontal, 94–95, 95n7, 101, 124, 169,
Portable power tool industry, 366 180, 230n6, 251
Post Office, 373, 374 vertical, 73, 82, 83, 95n7, 150, 164. See
Potassium Carbonate Association, 348 also Quality
Pot Still Malt Distillers Association, 345 Productivity. See Efficiency
Powered industrial truck and industrial Profitability, 3, 17–18, 19–20, 324, 326, 330
tractor industry, 364 and market structure as jointly endoge-
Prais, S. J., 98n11 nous variables. See Market structure,
Pratten, C. F., 122 and profitability/profits as jointly
Precast concrete goods industry, 402–403 endogenous variables
Precision File Association, 381 and price competition. See Price compe-
Precision Winding Association, 388 tition, and profitability/profits
Pressed Bowl Makers Association, 369 constructing the samples for the analy-
Pressed Brick Makers Association, 399 sis of, 294–297, 303–304
Pressed Steel, 217 data, 297–298
Price competition. See also Competition; econometric model for, 302–303, 309–
Intensity of price competition 312
Index 537

econometric results for, 313, 316–323 Ready-to-eat breakfast cereal industry,


measures of, 297–298, 303, 311n18 339
Profits. See Profitability Reddaway, W. B., 348
Proprietary Articles Trade Association, Reduced-form econometric models, 179–
350, 351 180, 229, 250–251, 274–275, 302, 311
Proprietary Association of Great Britain, Refractory goods industry, 398–399
350 Refrigerating machinery industry, 365–
Publishing (other than newspapers, 366
periodicals, and books) industry, 410– Refrigerator industry, 211–212, 216–221.
411 See also electrical appliance (domestic)
Pump industry, 361–362 industry
advertising, 211–212, 217, 218, 220, 221
Quality, 151, 164, 229. See also Product entry and exit, 216, 219
differentiation, vertical differences import penetration, 217, 219, 220
between firms, 73, 82, 83, 276. See also market size, 219
Asymmetric firms market structure, 211–212, 216, 217,
218–220, 221
R&D. See Research and development mergers, 219
R&D intensity, 76–77, 120–122, 239–240 price competition, 211–212, 216, 217,
and classification of industries, 120– 218, 220, 221
122, 239–240 prices, 216–217, 220
and collusive pricing, 79, 81, 82–83 restrictive agreement, 212
R&D-intensive industries, 5–7, 16, 35–36, retailers’ power, 217
76, 228–229, 239–240, 250, 283–285, Register of Restrictive Trading Agree-
326, 328 ments, 46–47, 51, 334, 335
constructing the samples of, 239–240, Registrar of Restrictive Trading Agree-
243–246, 251–252, 261–262 ments, 26, 27, 31, 33, 42, 47, 53
R&D-sales ratios, 76–77, 120–122, 239– reports, 24n, 40, 51–52, 336, 345, 356n
240 Reinforcement Conference, 383n
Radio and electronic components indus- Reiss, P. C., 93n6, 226
try, 374 Resale price maintenance, 22, 23, 24, 25,
Radio, radar, and electronic capital 28, 32, 34, 57, 58–59, 183, 189, 209, 333,
goods industry, 375 334
Railway Carriage and Wagon Building Lloyds’ Committee report on, 59, 335.
Association, 381 See also various industries
Railway carriage, wagon (railway freight Resale Prices Act (1964), 32, 58, 59n6
car), and tram (streetcar) industry, Research and development (R&D), 16,
381 121, 274
Railway Cast Bearings Association, 368 and marginal cost, 229
Raleigh, 380 and product quality, 82, 229
Rammer Manufacturers Association, 363 as sunk cost, 5–7, 91–92, 230
Random-effects model, 137–138, 188n, cooperative, 35
192, 255, 264–265, 312n20, 318 data, 120–122
Range Boiler Makers Association, 358 in advertising-intensive industries, 152,
Ransomes Sims and Jeferries, 361 162
Rapoport, J., 247 nontournament models of, 230
Rayon Staple Spinners and Doublers restrictions regarding, 35–36, 334
Association, 388 spillovers, 231, 231n7, 251, 256
Rayon Weaving Association, 389 Restrictive agreements, 1, 7, 21–28, 33,
Raw Fat Melters Association, 344 33n9, 333–336. See also Cartels; Collu-
Razor and blade industry, 382 sion; various industries
Ready-mixed concrete industry, 403 content of, 34–36, 46–47, 57–61
538 Index

Restrictive agreements (cont.) Roll Makers Association, 368


data sources on, 46–50 Roll (of iron and steel) industry, 368
effectiveness of, 37–39, 51–52, 56 Rolls Razor, 214, 215, 215n31, 217
for the export market, 24, 66–67, 334 Roofing felt industry, 403
nonregistration of, 28, 31–32, 47, 52–56 Rope, twine, and net industry, 392
registration of, 26–28, 29n7 Ross, D., 72, 149, 155, 326
Restrictive practices. See Collective dis- Rowley, C. K., 22n, 36, 38
crimination; Market sharing; Price- Royal Hand and Grocery Bags Associa-
fixing; Restrictive agreements tion, 408
Restrictive Practices Court, 21, 26, 27, Rubber and Thermoplastic Cable Manu-
28–31, 32, 33, 40, 41, 42, 47, 51, 53, facturers Association, 372
65n13, 88, 335 Rubber belting industry, 412
cases, 24n, 28–31, 29n6. See also various Rubber footwear industry, 398
industries Rubber Footwear Manufacturers Associ-
reports, 47, 335. See also various indus- ation, 398
tries Rubber hose and tubing industry, 412
Restrictive Trade Practices Act (1956), 1, Rubber industry, 411–412
7, 9, 10, 13, 21, 23–33, 43 Rubber Proofers Association, 412
effect on advertising intensity, 15–16, Rubber Thread Screw Association, 382
150, 178–179, 200–202, 205–211, 222, Rusk Manufacturers Association, 340
329–330
effect on competition, 1, 13, 21, 31, 32– Safe, lock, latch, and key industry, 385
33, 40–44, 51–52, 106, 108, 114, 130, Safety glass industry, 401
295n4, 298, 327n, 329, 336. See also Salmon Merchants Association of Great
various industries Britain, 341
effect on concentration, 14–16, 89, 128– Salop, S., 95n7
129, 138–146, 150, 178–179, 186–187, Salt Glazed Conduit Association, 399
188–198, 208–211, 222, 228–229, 263– Salt industry, 338
264, 265–274, 274–275, 329 Sand and gravel industry, 337
effect on firm/plant numbers, 17, 277, Sanitary towel (napkin) industry, 355
279, 307–309, 312–313, 314–315, 318, Sanitary earthenware industry, 400
319, 323–325 Sanitary ware (of fireclay, etc.) industry,
effect on innovation, 16, 228–229, 252– 399
253, 257–260, 274–276, 330 Sausage and cooked meat industry, 341
effect on prices, 41, 43n, 324–325 Sawyer, M. C., 107n
effect on profitability/profits, 17–18, Schankerman, M., 247, 282n
43n, 277, 278, 279, 307–309, 313, 316– Scheinkman, J. A., 73
322, 323–325, 329 Scherer, F. M., 72, 91, 149, 155, 226, 238n,
implementation of, 13, 21, 28–32 326
provisions of, 21, 23–24, 25–26, 31–32 Schmalensee, R., 152n2
Restrictive Trade Practices Act (1968), 31, Schnee, J., 247
32, 41 Schott, K., 247, 249
Retread Manufacturers Association, 411 Schumpeter, J. A., 223
Rey, P., 224 Schwartz, N. L., 232
Road Roller Manufacturers Association, Schweppes, 345
363 Science Policy Research Unit (SPRU), 240
Roadstone Producers Advisory Com- innovations database, 240–250, 252,
mittee, 337 253, 259
Robinson, W. T., 88n, 193n22 Scientific and industrial instruments and
Robson, M., 241, 242 systems industry, 370
Rod Rollers Association, 359 Scottish Association of Cooked Meat
Rolling mill industry, 366 Manufacturers, 341
Index 539

Scottish Association of Edible Fat Semicollusion, 155, 224n, 233


Melters, 344 Seneca, J. J., 74
Scottish Association of Manufacturing Setup cost, 5, 6, 91
Coppersmiths, 387 and collusion, 78, 79, 80, 83, 128
Scottish Association of Master Bakers, and concentration, 79, 128, 139
339 measures of, 75–76, 114–116, 133n,
Scottish Firebrick Association, 398 254n, 299
Scottish Fireclay Pipe Association, 399 Shaked, A., 100, 101n15
Scottish Glass Merchants and Glaziers Shapley, L., 101n14
Association, 401 Shaw, R. W., 213, 214
Scottish Hot Road Binder Manufacturers Sheffield District Annealers and Heat
and Spraying Association, 404 Treaters Association, 387
Scottish House Furnishers Association, Sheffield Ganister and Compo. Associa-
406 tion, 398
Scottish Joinery and Door Manufacturers Shell, 347, 348
Association, 405 Ship and Boat Builders National Federa-
Scottish Net Manufacturers Association, tion, 378
392 Shipbuilding industry, 378
Scottish Provision Trade Association, 340 Ship repairing industry, 378
Scottish Road Emulsion Association, 404 Ships Ordinary Sidelights Association,
Scottish Soda Crystal Manufacturers 387
Association, 348 Shive Manufacturers Association, 407
Scottish Surgical Instrument Manufac- Shoe Tip Association, 398
turers Association, 369 Shop and office fitting industry, 406
Scottish Woollen Spinners Association, Shopfront Moulding Manufacturers
390 Association, 406
Screw Stopper Makers Association, 412 Short Saw and Crosscut Saw Associa-
Screw Thread Tool Manufacturers Asso- tion, 381
ciation, 381 Shubik, M., 101n14, 102n16
Scythe, Sickle and Hook Manufacturers Silberston, A., 137
Association, 381 Silk and Rayon Trade Protection Society,
Seade, J., 94 392
Secondary battery industry, 290–294, Silk Throwsters Association, 389
376–377 Silk Trade Employers Association, 389
demand growth, 292 Simon, H. A., 98n11
entry barriers, 291, 293 Simons, K., 226
market structure, 290, 291, 292–293 Skip Plant Association, 365
Monopolies Commission inquiry, 290, Slade, M., 149
291–292, 376–377 Slag Wool Association, 404
National Board for Prices and Incomes Slate industry, 337
report, 292 Sleuwaegen, L., 88
price competition, 290–292, 377 Small Offset Association, 411
prices, 292–293 Smith, 376
profits, 292–293 Soap and detergent industry, 351–352
restrictive agreements, 290–292, 377 Society of British Gas Industries, 386
Restrictive Practices Court case, 291, Society of British Ink Manufacturers,
377 354
Secretary of State for Prices and Con- Society of British Paint Manufacturers,
sumer Protection (SSPCP), 33n8 351
Self-Raising Flour Association, 344 Society of Crepe Paper Makers , 409
Self-rising flour industry, 344 Society of Laundry Engineers and Allied
Selten, R., 87, 150, 277, 280, 319 Trades, 367
540 Index

Society of Motor Manufacturers and Stockinette Manufacturers Association,


Traders, 379 392
Society of Parliamentary Agents, 411 Stockless Anchor Association, 387
Society of Photo Printers, 411 Stone mining and quarrying industry,
Soft drink industry, 345 336–337
Soft Hemp and Tow Spinners Associa- Stourbridge Firebrick Association,
tion, 389 398
South Eastern Brick and Tile Federation, Stourbridge Glass Manufacturers Associ-
399 ation, 402
Southern Lime Association, 403 Structure-conduct-performance para-
South of England Hat Manufacturers digm, 3, 93, 93n5, 95n7
Federation, 397 Subgame perfect equilibrium, 103, 151,
South of Scotland Electricity Board, 375 167, 229
South Wales and Mon. Sanitary Pipe Suarez, F. F., 226
Association, 399 Sugar confectionery industry, 343
South West of England Lime Associa- Sugar industry, 342
tion, 403 Summation Meter Manufacturers Associ-
Space-heating, ventilating, and air- ation, 370
conditioning equipment industry, Sunk costs, 5–7, 14, 15, 18, 87, 96, 115,
366 146, 280, 324, 327–328
Spectacles (eyeglasses) and lenses indus- endogenous, 5–7, 91–92. See also Adver-
try, 370 tising; Research and development
Spence, M., 100 exogenous, 5–6, 91–92. See also Setup
Spencer, B. J., 100n costs
Spinning and doubling (on the cotton Superphosphate Manufacturers Associa-
and flax systems) industry, 29–30, tion, 353
388–389 Surgical Appliance Manufacturers Asso-
Spirit distilling and compounding indus- ciation, 369
try, 345–346 Surgical Dressings Manufacturers Asso-
Sports equipment industry, 413 ciation, 354
Spring and Interior Springing Associa- Surgical and medical bandages industry,
tion, 385 354–355
Springs industry, 385–386 Suslow, V., 74n17
Stainless Steel Bar Products Association, Sutton, C. J., 213, 214
382 Sutton, J., 3, 5, 11, 87, 93n6, 97, 98n10,
Starch and glucose industry, 344 98n11, 100, 101n14, 101n15, 106,
Stationers’ goods industry, 413 114, 150, 154, 165, 166n10, 167, 193,
Stationers Proprietary Articles Trade 219n34, 226, 230n5, 230n6, 277,
Association, 409, 413 313n23, 319
Stationery industry, 409 Swann, D., 21, 22n, 24n, 27, 28, 31, 36,
Steam Turbine and Associated Plant 37, 38, 39, 41–43, 43n, 51, 55, 81n24,
Manufacturers, 371 146, 228, 285–292, 329, 330, 335, 337,
Steel Hinge Makers Association, 387 339, 355, 357, 358, 359, 371, 372, 373,
Steel (general) industry, 355–356 377, 383, 385, 391, 393, 399, 400, 401,
Steel tube industry, 357 402, 412
Steelwork and ironwork industry, 367– Switchboard Cable Association, 372
368 Switchgear industry, 371
Steel Works Plant Association, 366 Symeonidis, G., 33n8, 72, 73, 75n19, 76,
Stewards and Lloyds, 357 80, 101, 106n19, 114n, 131, 152n3, 165,
Stiglitz, J., 95n7, 230n5 167, 168, 225
Stock Brick Manufacturers Association, Synthetic Cordage Manufacturers Asso-
399 ciation, 392
Index 541

Synthetic resins and plastics materials Twine Manufacturers Association, 392


industry, 352 Twist Drill Association, 381
Synthetic rubber industry, 352 Typewriter and Allied Trades Federation
of Great Britain, 364, 375
Tank and Industrial Plant Association, Tyre Manufacturers Conference, 411
368
Tape Manufacturers Association, 393 Unilever, 351
Technological opportunity, 6, 77, 82, 85, U.K. Glycerine Producers Association,
92, 225, 231, 251, 253, 259 352
Telegraph and telephone apparatus and U.K. White Lead Convention, 352
equipment industry, 35, 36, 82, 373– Umbrella Components Association, 387
374 Underwood Products Association, 407
Telephone Cable Makers Association, Urata, S., 278
372 Utterbach, J. M., 226
Textile finishing industry, 394–395 Utton, M. A., 215, 219n35, 342, 346, 373,
Textile industries, miscellaneous, 395 374, 390, 391
Textile machinery industry, 363
Thisse, J.-F., 95n8, 150, 281 Valve industry, 362
Thomas, C. J., 350 Van Reenen, J., 225, 248, 255n
Thomas, G., 241, 244, 249 Vegetable and animal oil and fat indus-
Tie Manufacturers Association, 398 try, 344
Tile (of earthenware) industry, 65n13, Ventile Fabrics Association of Great
400 Britain, 389
Timber industry, 405 Vickers, J., 96
Tinbox Manufacturers Association, 384 Vives, X., 100
Tin industry, 360 Vitreous Enamellers Association, 387
Tire industry, 411–412 Vulcanised Fibre and Leatheroid Associ-
Tirole, J., 73, 92 ation, 409
Tobacco industry, 346 Vuong, Q., 149
Tobacco Trade Association, 346
Toilet preparations industry, 351 Wagner, S., 247
Townsend, H., 382 Waldman, D. M., 124
Townsend, J., 241, 242, 244, 249 Wallpaper industry, 409
Toys and games industry, 412 Wallpaper Manufacturers and Employ-
Traced Art Needlework Manufacturers ers Association, 409
Association, 394 Walshe, G., 215, 219n35, 342, 346, 373,
Trade associations, 22, 27, 33, 34, 36, 59, 374, 390, 391
81. See also Political and Economic Ward, T. S., 183
Planning, survey of trade associations; Washing machine industry, 211–212,
various industries 213–216. See also electrical appliance
Trade union density, 301–302 (domestic) industry
Trade union power, 301–302 advertising, 205, 211–212, 213, 214, 215
Trailer and caravan industry, 379 distribution, 214, 215
Transformer Makers Association, 286 entry and exit, 214, 215
Trawl Twine Manufacturers Association, import penetration, 215
392 market size, 215, 215n32
Triplex, 401 market structure, 211–212, 213, 214,
Trivedi, P. K., 255, 255n 215, 215n31
Troup, G. W., 361 mergers, 215, 219n35
Tube Investments, 357, 380 price competition, 211–212, 213–214
Tubular Frame Saw Association, 381 prices, 214
Tuyere Makers Association, 368 restrictive agreement, 212
542 Index

Watch industry, 369 Wrought Hollow-ware Trade Employers


Waterson, M., 107n, 311 Association, 384
Water-Tube Boilermakers Association, Wyatt, S., 241, 244, 249
367
Weatherproof outerwear industry, 396 Yamawaki, H., 88
Weaving (of cotton, linen and man-made Yamey, B. S., 58, 59
fibers) industry, 389–390 Yarn Spinners Association, 29–30, 388
Welded Tool Manufacturers Association, Yarrow, G. K., 150, 224
381 Yorkshire Firebrick Association, 398
Welding and flame-cutting equipment
industry, 361 Zinc industry, 360
Well Drillers Association, 363 Ziss, S., 150, 224
Welsh Silica Association, 398
Wensleydale Cheese Joint Conference,
341
Wheeled tractor industry, 378
Whiskey industry, 346
White, H., 138, 188, 312
White Oils Association, 348
Wholesale and Retail Bakers of Scotland
Association, 339
Wholesale Confectioners Alliance of
Great Britain and Northern Ireland,
343
Wilberforce, R. O., 24n
Williams, B. R., 399, 400
Wilson, T., 88, 228
Window Holland Association, 389
Wire and wire manufactures industry,
383
Wire Rope Manufacturers Association,
383
Women’s and girls’ tailored outerwear
industry, 396–397
Wood chipboard industry, 405
Wooden container and basket industry,
407
Wood Handle Manufacturers Council,
407
Wood industries, miscellaneous, 407
Woodville District Firebrick and Fireclay
Association, 398
Woolen and worsted industry, 390–
391
Wool Carbonisers and Scourers Federa-
tion, 390
Woolcombers Association, 390
Woollen and Worsted Manufacturers
Association, 391
Worsted Spinners Federation, 390
Wright, D. M., 89–90, 277–278, 329
Wright, N. J., 132

Das könnte Ihnen auch gefallen